🗺️1202map - amoon.world🌙

# final exam using [incorporating llm skills in final exam cld](https://claude.ai/chat/27e13038-3f0b-43aa-9dce-273f755aa87d), i proposed to include llm for taking finals based on [[TEPEI Referee Report Assignment.pdf]] # r11 [[2.💻Input2Algorithm2Output Module]] [[3.⏰Time Allocation Module: Preparation and Execution Framework]] | Component | Input with file extensions | Algorithm | Output | goal | prompt to answer | | --------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- | | **1. Review Class by 🧙prof.moshe ben-akiva by 🙋‍♀️angie moon (TA)** | • 🗣️ Lecture transcripts by moshe (min. 2) ending with otter_ai.txt • 📝Lecture notes/slides (min. 2) ending with .pdf • 🖼️Diagrams from lecture materials ending with .png [[1.202 - Demand Modeling_otter_ai (2).txt]] | 1. Analyze 🗣️transcripts and 📝notes to identify 🔑key concepts 2. Project information onto two axes: a. "🧙Moshe's Recipe" (theoretical concepts) b. "🙋‍♀️Angie's Recipe" (programming implementation) 3. Prioritize concepts with visual representations 4. Extract moshe's wisdom statements | • Three 🔑key demand modeling concepts with 🖼️diagrams • Three 🧙moshe's wisdom points from "🧙Moshe's Recipe" • Code examples showing practical application from "🙋‍♀️Angie's Recipe" | 🚨infer moshe's desired knowledge state together | 🚨answer what is moshe's desired knowledge state? | | **2. Preview Case Study** | • Outcomes from 1. Review Class • Case study materials ending with .pdf that includes 💻 in filename ![[Mixture_SM_RandomCoeff.py]] [[💻CaseStudy4_23.pdf]], [[💻CS4_23_Brief.pdf]] | 1. Identify bridges between lecture concepts and case requirements 2. Create concise connections between concepts and applications 3. Develop guidance based on theoretical understanding | • Three concise bridging sentences that: a. Connect theory to case requirements b. Acknowledge different knowledge states c. Motivate with practical relevance | 🚨🚨 introduce application context | 🚨🚨answer how can students apply theoretical concepts in the case study? | | **3. Check-in Task** | • Outcome from 1. Review Class and 2. Preview Case Study • Previous recitation transcript (optional) | 1. Create real-world hypothesis scenarios 2. Follow structure: "[Stakeholder]'s hypothesis: [Subject] are more/less likely to [action] when [condition]" 3. Design task requiring theoretical understanding and coding 4. Ensure task reveals knowledge gaps | • Three hypothesis-driven prompts that: a. Incorporate key concepts b. Require programming c. Complete within 15 minutes d. Evoke 😲surprise from misconceptions | 🚨🚨🚨let students infer their knowledge state | 🚨🚨🚨answer what are three prompts i should give to students to help students learn their knowledge state? | | **4. Data Collection** | • Outcomes from Tasks 1-3 • Previous data collection results | 1. Identify knowledge gaps from check-in task 2. Design exclusive in-person data collection method 3. Structure collection to measure knowledge gain 4. Focus on qualitative understanding over self-reporting | • One data collection method that: a. Provides immediate value to attendees b. Completes within 10 minutes c. Yields actionable insights d. Prioritizes conceptual understanding | 🚨🚨🚨🚨angie learns students' knowledge state | 🚨🚨🚨🚨answer how can angie bridge students' knowledge state with moshe's desire with programming in case study? | # r10 using [gpt](https://chatgpt.com/c/68025acf-7220-8002-8242-1d922e2346db) | Component | Input with file extensions | Algorithm | Output | goal | prompt to answer | | --------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- | | **1. Review Class by 🧙prof.moshe ben-akiva by 🙋‍♀️angie moon (TA)** | • 🗣️ Lecture transcripts by moshe (min. 2) ending with otter_ai.txt • 📝Lecture notes/slides (min. 2) ending with .pdf • 🖼️Diagrams from lecture materials ending with .png [[🗣️ Demand Modeling mixture nested logit_otter_ai.txt]], [[🗣️Demand Modeling nested logit_otter_ai.txt]] [[📝s25_Lecture18_NestedLogit.pdf]], [[📝s25_Lecture19_MEVmodels.pdf]] | 1. Analyze 🗣️transcripts and 📝notes to identify 🔑key concepts 2. Project information onto two axes: a. "🧙Moshe's Recipe" (theoretical concepts) b. "🙋‍♀️Angie's Recipe" (programming implementation) 3. Prioritize concepts with visual representations 4. Extract moshe's wisdom statements | • Three 🔑key demand modeling concepts with 🖼️diagrams • Three 🧙moshe's wisdom points from "🧙Moshe's Recipe" • Code examples showing practical application from "🙋‍♀️Angie's Recipe" | 🚨infer moshe's desired knowledge state together | 🚨answer what is moshe's desired knowledge state? | | **2. Preview Case Study** | • Outcomes from 1. Review Class • Case study materials ending with .pdf that includes 💻 in filename [[💻CaseStudy4_23.pdf]], [[💻CS4_23_Brief.pdf]] | 1. Identify bridges between lecture concepts and case requirements 2. Create concise connections between concepts and applications 3. Develop guidance based on theoretical understanding | • Three concise bridging sentences that: a. Connect theory to case requirements b. Acknowledge different knowledge states c. Motivate with practical relevance | 🚨🚨 introduce application context | 🚨🚨answer how can students apply theoretical concepts in the case study? | | **3. Check-in Task** | • Outcome from 1. Review Class and 2. Preview Case Study • Previous recitation transcript (optional) | 1. Create real-world hypothesis scenarios 2. Follow structure: "[Stakeholder]'s hypothesis: [Subject] are more/less likely to [action] when [condition]" 3. Design task requiring theoretical understanding and coding 4. Ensure task reveals knowledge gaps | • Three hypothesis-driven prompts that: a. Incorporate key concepts b. Require programming c. Complete within 15 minutes d. Evoke 😲surprise from misconceptions | 🚨🚨🚨let students infer their knowledge state | 🚨🚨🚨answer what are three prompts i should give to students to help students learn their knowledge state? | | **4. Data Collection** | • Outcomes from Tasks 1-3 • Previous data collection results | 1. Identify knowledge gaps from check-in task 2. Design exclusive in-person data collection method 3. Structure collection to measure knowledge gain 4. Focus on qualitative understanding over self-reporting | • One data collection method that: a. Provides immediate value to attendees b. Completes within 10 minutes c. Yields actionable insights d. Prioritizes conceptual understanding | 🚨🚨🚨🚨angie learns students' knowledge state | 🚨🚨🚨🚨answer how can angie bridge students' knowledge state with moshe's desire with programming in case study? | # r9 [[cs3]] | Component | Input with file extensions | Algorithm | Output | goal | prompt to answer | | --------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- | | **1. Review Class by 🧙prof.moshe ben-akiva by 🙋‍♀️angie moon (TA)** | • 🗣️ Lecture transcripts by moshe (min. 2) ending with otter_ai.txt • 📝Lecture notes/slides (min. 2) ending with .pdf • 🖼️Diagrams from lecture materials ending with .png | 1. Analyze 🗣️transcripts and 📝notes to identify 🔑key concepts (what moshe - speaker - emphasizes) 2. Project information onto two axes: a. "🧙Moshe's Recipe" (theoretical concepts) b. "🙋‍♀️Angie's Recipe" (programming implementation) 3. Prioritize concepts with visual representations 4. Extract moshe's wisdom statements | • Three 🔑key demand modeling concepts with 🖼️diagrams • Three 🧙moshe's wisdom points from "🧙Moshe's Recipe" • Code examples showing practical application from "🙋‍♀️Angie's Recipe" | 🚨infer moshe's desired knowledge state together | 🚨answer what is moshe's desired knowledge state? | | **2. Check-in Task** | • Outcome from 1. Review Class, name this `out1` • Previous recitation transcript (optional) | 1. Create real-world hypothesis scenarios building on `out1` 2. Follow structure: "[Stakeholder]'s hypothesis: [Subject] are more/less likely to [action] when [condition]" 3. Design task requiring both theoretical understanding and coding 4. Ensure task reveals knowledge gaps | • Three hypothesis-driven prompts that: a. Incorporate key concepts b. Require programming c. Complete within 15 minutes d. Evoke 😲surprise from misconceptions | 🚨🚨let students infer their knowledge state | 🚨🚨answer what are three prompts i should give to students to help students learn their knowledge state? | | **3. Data Collection** | • Outcomes from Tasks 1-2, name this `out1`, `out2` • Previous data collection results | 1. Identify potential knowledge gaps from check-in task 2. Design exclusive in-person data collection method 3. Structure collection to measure knowledge gain 4. Focus on qualitative understanding over self-reporting | • One data collection method that: a. Provides immediate value to attendees b. Completes within 10 minutes c. Yields actionable insights d. Prioritizes conceptual understanding by analyzing 😲surprise from misconceptions | 🚨🚨🚨angie learns students' knowledge state | 🚨🚨🚨answer what should angie focus on when listening to student's answer to learn their knowledge state? | | **4. Preview Case Study** | • Outcomes from Tasks 1-3 • Case study materials | 1. Identify bridges between current knowledge and case requirements 2. Create concise connections between concepts, tasks, and applications 3. Develop guidance based on observed knowledge states | • Three concise bridging sentences that: a. Connect theory to case requirements b. Acknowledge different knowledge states c. Motivate with practical relevance | 🚨🚨🚨🚨 angie helps students reach to moshe's desire from each of your knowledge state | 🚨🚨🚨🚨answer how can angie bridge students' knowledge state with moshe's desire with programming in case study? | # r8 | each section's desire | input | algorithm | output | | ---------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- | | 1. review class 🚨goal1: infer moshe's desired knowledge state together | 🗣️ transcript1 and 2 ![[1.202 - Demand Modeling model specification.txt]] ![[1.202 demand modeling - aggregated forecasting_otter_ai.txt]] 📝lecture note 1,2,3 ![[1.202_s25_Lecture15a_Forecasting.pdf]] ![[1.202_s25_Lecture16_StatisticalTests.pdf]] | project 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to my three 🤜action space in the following three steps 1. imagine an information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3. 2. imagine two axes: 🧙‍♂️🍱 moshe's recipe of cuisines, 💻🍱 angie's recipe for programming. 3. project the information space from 1 to the axes in 2. outcome of 3 should provide the most efficient review of what moshe (instructor) intended students to learn about demand modeling in a format of teaching programming. 🧙‍♂️🍱 moshe's recipe of cuisines consists of💡three key ideas (with higher preference on the concepts with diagram from 📝lecture note) wrapped in a 🧙‍♂️ three moshe's wisdom angie found interesting (moshe wants to teach us to be a chef, not instill recipe) 💻🍱 angie's recipe for programming shows (visualizes) with code the concepts explained in class | 🚨task1: answer what is moshe's desired knowledge state? | | 2. check-in task 🚨🚨goal2: let students infer their knowledge state | | building on outcome of 🚨task1, design three prompts that students can digest 🧙‍♂️🍱 moshe's recipe of cuisines and implement with 💻🍱 angie's recipe for programming. before - 👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am - 👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain - 🏢city council's hypothesis: citizen's car ownership are less likely to support 🚸pedestrianization than non-car owners - 🌙amoon's hypothesis: 👨‍🎓student's 📚assignment load and 😋hungriness affect computer lab attendance | 🚨🚨task2: what are three prompts i should give to students to help students learn their knowledge state? | | 3. discuss to collect data 🚨🚨🚨goal3: angie learns students' knowledge state | 🤩angie's inference and allocation 1. desire: infer moshe's desired knowledge state of students (review class), infer students current knowledge state (check-in+data collection), prepare students to bridge that gap by programming case study () 2.output measure: by your grades (customer: moshe) and course evaluation (customer: students) 3. reward: quality of case study 1~5 submission | building on outcome of 🚨task1, 🚨🚨task2, 🤩angie's inference and allocation, design one data i should collect, given my goal to maximize average delta K (knowledge gain) from student participating the recitation. last class, I collected self-reported participation. | 🚨🚨🚨task3: what should angie focus on when listening to student's answer to learn their knowledge state? | | 4. preview case study 🚨🚨🚨🚨goal4: angie helps students reach to moshe's desire from each of your knowledge state | [[CaseStudy2_22_Solutions.pdf]] [[Midterm_23_22_18_16_sol 1.pdf]] | building on outcome of 🚨task1, 🚨🚨task2, 🚨🚨🚨task3, explain in three sentence how i should curate case study to students to motivate them. | 🚨🚨🚨🚨task4: how can angie bridge students' knowledge state with moshe's desire with programming in case study? | # midterm prep - relation between lagged variable, autocorrelation, endogeneity - simultaneous equations modeling supply and demand | | lessons | detail | | --------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | r6 time series iv | 🍱1. Time Series Autocorrelation & Its Consequences | Time series data introduces the challenge of autocorrelation - where error terms are correlated over time. Under autocorrelation: - OLS estimators remain unbiased and consistent but become inefficient - Standard errors from normal OLS calculations are incorrect - With lagged dependent variables, OLS becomes inconsistent and biased | | | 🍱2. Endogeneity & The Need for Instrumental Variables | Endogeneity occurs when regressors are correlated with error terms, causing OLS to be biased and inconsistent. Key causes: - Omitted variables: Missing important explanatory variables - Measurement errors: Noisy measurement of explanatory variables (Q1) - Simultaneity: Explanatory variables determined jointly with dependent variable - Lagged dependent variables with autocorrelated errors (nonzero rho) - Self-selection: Non-random assignment of treatments | | | 🍱3. Simultaneous Equations Models & Identification | In simultaneous equations, variables influence each other reciprocally. For example, in supply and demand: The key insight is that identification requires: Different exogenous variables in each equation (X₁ ≠ X₂) Using variables that shift one curve but not the other Two-Stage Least Squares estimation of the structural equations | | r5 violation of ols, gls | 🍱1: Violation of OLS Assumptions and Consequences | when OLS assumptions are violated 1. No Perfect Multicollinearity (rank(X) = K) → Non-existence of estimates 2. Strict Exogeneity (E(ε\|X) = 0) → Biased estimates 3. Spherical Error Variance (E(εε'\|X) = σ²I) → Inefficient but unbiased estimates 3.1 Homoscedasticity (E(ε²ₙ\|X) = σ² > 0) 3.2 No autocorrelation (E(εₙεₙ'\|X) = 0) 4. Normality (ε\|X ~ N(0,σ²I)) → Affects distribution-based inference | | | 🍱2: Weighted Least Squares (WLS) for Heteroscedasticity | When heteroscedasticity is present (variance of errors is not constant), WLS transforms the model: - Original model: yₙ = x'ₙβ + εₙ with Var(εₙ\|x) = σ²ₙ - We weight each observation by 1/σₙ to create: yₙ/σₙ = (xₙ/σₙ)'β + εₙ/σₙ - The transformed model has homoscedastic errors: Var(εₙ/σₙ) = 1 - Two approaches for implementation: 1. Run OLS on transformed variables (yₙ/σₙ and xₙ/σₙ) 2. Use weighted least squares on original variables with weight 1/σ²ₙ | | | 🍱3:Generalized Least Squares for Non-Spherical Errors | For general non-spherical errors (E(εε'\|X) = σ²V), GLS provides a comprehensive solution: Decompose V⁻¹ = C'C and transform the model: Y* = CY, X* = CX, ε* = Cε Transformed model satisfies assumptions I-III GLS estimator: β̂ = (X'V⁻¹X)⁻¹X'V⁻¹Y Variance: Var(β̂\|X) = σ²(X'V⁻¹X)⁻¹ When V is unknown, Feasible GLS can be used: Run OLS to get β̂ₒₗₛ Use residuals to estimate V̂ Apply GLS with V̂ | | r4 | 🍱1. Statistical Properties of Linear Regression | four key assumptions that make regression valid: 1: No perfect multicollinearity - The design matrix must be full rank, meaning no exact linear relationship among independent variables 2: Strict exogeneity - The expected value of the error term given X is zero 3: Spherical errors - Homoscedasticity (constant variance) and no autocorrelation 4: Normality - Error terms are normally distributed | | | 🍱2. Goodness of Fit Assessment | R² is the correlation squared between observed and fitted values Adjusted R² penalizes for the number of parameters to avoid overfitting R² is most useful for comparing models with identical data and dependent variables The value of R² alone doesn't determine if a model is "good" or "bad" | | | 🍱3. Hypothesis Testing Framework | Proper hypothesis testing in regression requires: Understanding when t-tests are appropriate for individual coefficients Using F-tests for testing multiple restrictions Recognizing the costs of Type I vs Type II errors when making model decisions Prioritizing theoretical meaning over statistical significance | | | 🧙‍♂️ | 1. "Domain knowledge trumps statistics" most important criteria for model evaluation is whether the coefficients make sense based on domain knowledge (demand and price). Statistical tests are secondary to theoretical validity. 2. "Models are tools for analysis, not just prediction" The purpose of demand modeling is to perform analysis through manipulation - understanding how the dependent variable changes when we intervene on independent variables, not just fitting data. 3. "Beware the P-hacking trap" warns against the practice of specification searching to find variables with high t-statistics. Statistical significance alone doesn't justify inclusion or exclusion of variables. | | r3 | Type I vs Type II Error Trade-off | Type I (α): Probability of rejecting H0 when true, false positive Type II (β): Probability of accepting H0 when false, false negative Key trade-off: Reducing α increases β for given sample size Can only reduce both by increasing sample size | | | Sample Size Determination Approaches: | set allowable error and confidence level in advance ES increases then power increases consider cost constraints and available budget Different approaches for different data types (proportions vs means) but this is only one type of estimation errors (responsive and non-responsive error) | | | Stratified Sampling three steps (1200 to 900 to 700): | Worthwhile when population variance differs by strata Allows for more efficient allocation of sampling resources low, medium, high (300, 600, 300) -(?)-> (300, 300, 300) -(?)-> () | | | 🧙‍♂️1 consider prior probability of null hypothesis being true Cost of Type I vs Type II errors Context of the decision being made 🧙‍♂️2 Allocate resources to where you know the least - When designing sampling strategies: Focus data collection on segments with high variability Use prior knowledge to optimize sample allocation Be strategic rather than mechanistic in sampling approach | | | r2 | | | | r1 | | | --- # r6 # ⏰time allocation | Planning Component | One-Sentence Instruction | Example from Your Framework | Time Allocation | | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | 1. Review Class | Synthesize lecture transcripts and notes into a structured framework presenting three key theoretical concepts with visual representations (Moshe's recipe), three instructor insights (Moshe's wisdom), and corresponding programming implementations (Angie's recipe). | "Project information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to two axes: 🧙‍♂️🍱 moshe's recipe of cuisines (three key ideas with diagrams wrapped in three moshe's wisdom) and 💻🍱 angie's recipe for programming (code visualizations of concepts)." | **Prep:** 2 hours **Class:** 15 minutes | | 2. Check-in Task | Design a hypothesis-driven real-world problem that requires students to apply the week's key concepts through programming implementation, following the pattern of your examples. | "Design a task like '👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am' or '👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain' that students can implement using programming." | **Prep:** 2 hours **Class:** 15 minutes | | 3. Data Collection | Design and implement an exclusive in-person data collection experience that provides immediate value to attendees while measuring specific knowledge gains beyond self-reported participation. | "Design one data collection method to maximize average delta K (knowledge gain) from students participating in the recitation, improving upon previously collected self-reported participation." | **Prep:** 4 hour - grading to give feedback (⭐️will prioritize who attend recitation) **Class:** 10 minutes | | 4. Preview Case Study | Create three concise sentences that explicitly connect the theoretical concepts from lectures, the practical skills from the check-in task, and the relevant applications in the course case study. | "Building on review class and check-in task, explain in three sentences how case study connects with the outcomes of previous components." | **Prep:** 2 hours **Class:** 15 minutes | | | **TOTAL** | | **Prep:** 10 hours **Class:** 55 minutes | # 🤜input2algorithm2output | each section's desire | input | algorithm | output | | ---------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- | | 1. review class 🚨goal1: infer moshe's desired knowledge state together | 🗣️ transcript1 and 2 [[1.202 - Demand Modeling Inst.Var_otter_ai.txt]] [[1.202 - Demand Modeling Time Series_otter_ai.txt]] 📝lecture note 1,2,3 [[1.202_s25_Lecture11_TimeSeriesAnalysis.pdf]] [[1.202_s25_Lecture12_IV.pdf]] | project 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to my three 🤜action space in the following three steps 1. imagine an information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3. 2. imagine two axes: 🧙‍♂️🍱 moshe's recipe of cuisines, 💻🍱 angie's recipe for programming. 3. project the information space from 1 to the axes in 2. outcome of 3 should provide the most efficient review of what moshe (instructor) intended students to learn about demand modeling in a format of teaching programming. 🧙‍♂️🍱 moshe's recipe of cuisines consists of💡three key ideas (with higher preference on the concepts with diagram from 📝lecture note) wrapped in a 🧙‍♂️ three moshe's wisdom angie found interesting (moshe wants to teach us to be a chef, not instill recipe) 💻🍱 angie's recipe for programming shows (visualizes) with code the concepts explained in class | 🚨task1: answer what is moshe's desired knowledge state? | | 2. check-in task 🚨🚨goal2: let students infer their knowledge state | | building on outcome of 🚨task1, design a task that students can digest 🧙‍♂️🍱 moshe's recipe of cuisines and implement with 💻🍱 angie's recipe for programming. before - 👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am - 👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain - 🏢city council's hypothesis: citizen's car ownership are less likely to support 🚸pedestrianization than non-car owners - 🌙amoon's hypothesis: 👨‍🎓student's 📚assignment load and 😋hungriness affect computer lab attendance | 🚨🚨task2: what are three prompts i should give to students to help students learn their knowledge state? | | 3. discuss to collect data 🚨🚨🚨goal3: angie learns students' knowledge state | 🤩angie's inference and allocation 1. desire: infer moshe's desired knowledge state of students (review class), infer students current knowledge state (check-in+data collection), prepare students to bridge that gap by programming case study () 2.output measure: by your grades (customer: moshe) and course evaluation (customer: students) 3. reward: quality of case study 1~5 submission | building on outcome of 🚨task1, 🚨🚨task2, 🤩angie's inference and allocation, design one data i should collect, given my goal to maximize average delta K (knowledge gain) from student participating the recitation. last class, I collected self-reported participation. | 🚨🚨🚨task3: what should angie focus on when listening to student's answer to learn their knowledge state? | | 4. preview case study 🚨🚨🚨🚨goal4: angie helps students reach to moshe's desire from each of your knowledge state | [[CaseStudy2_22_Solutions.pdf]] [[Midterm_23_22_18_16_sol 1.pdf]] | building on outcome of 🚨task1, 🚨🚨task2, 🚨🚨🚨task3, explain in three sentence how i should curate case study to students to motivate them. | 🚨🚨🚨🚨task4: how can angie bridge students' knowledge state with moshe's desire with programming in case study? | # r5 2025-03-07 # ⏰time allocation | Planning Component | One-Sentence Instruction | Example from Your Framework | Time Allocation | | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | 1. Review Class | Synthesize lecture transcripts and notes into a structured framework presenting three key theoretical concepts with visual representations (Moshe's recipe), three instructor insights (Moshe's wisdom), and corresponding programming implementations (Angie's recipe). | "Project information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to two axes: 🧙‍♂️🍱 moshe's recipe of cuisines (three key ideas with diagrams wrapped in three moshe's wisdom) and 💻🍱 angie's recipe for programming (code visualizations of concepts)." | **Prep:** 2 hours **Class:** 15 minutes | | 2. Check-in Task | Design a hypothesis-driven real-world problem that requires students to apply the week's key concepts through programming implementation, following the pattern of your examples. | "Design a task like '👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am' or '👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain' that students can implement using programming." | **Prep:** 2 hours **Class:** 15 minutes | | 3. Data Collection | Design and implement an exclusive in-person data collection experience that provides immediate value to attendees while measuring specific knowledge gains beyond self-reported participation. | "Design one data collection method to maximize average delta K (knowledge gain) from students participating in the recitation, improving upon previously collected self-reported participation." | **Prep:** 4 hour - grading to give feedback (⭐️will prioritize who attend recitation) **Class:** 10 minutes | | 4. Preview Case Study | Create three concise sentences that explicitly connect the theoretical concepts from lectures, the practical skills from the check-in task, and the relevant applications in the course case study. | "Building on review class and check-in task, explain in three sentences how case study connects with the outcomes of previous components." | **Prep:** 2 hours **Class:** 15 minutes | | | **TOTAL** | | **Prep:** 10 hours **Class:** 55 minutes | # 🤜input2algorithm2output | each section's desire | input | algorithm | output | | ----------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- | | 1. review class 🚨goal1: infer moshe's desired knowledge state together | 🗣️ transcript1 and 2 [[1.202 - Demand Modeling gls.txt] [[1.202 - Demand modeling ols violation.txt]] 📝lecture note 1,2,3 [[1.202_s25_Lecture9_ViolationofOLSAssumptions.pdf]] [[1.202_s25_Lecture10_GLS.pdf]] | project 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to my three 🤜action space in the following three steps 1. imagine an information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3. 2. imagine two axes: 🧙‍♂️🍱 moshe's recipe of cuisines, 💻🍱 angie's recipe for programming. 3. project the information space from 1 to the axes in 2. outcome of 3 should provide the most efficient review of what moshe (instructor) intended students to learn about demand modeling in a format of teaching programming. 🧙‍♂️🍱 moshe's recipe of cuisines consists of💡three key ideas (with higher preference on the concepts with diagram from 📝lecture note) wrapped in a 🧙‍♂️ three moshe's wisdom angie found interesting (moshe wants to teach us to be a chef, not instill recipe) 💻🍱 angie's recipe for programming shows (visualizes) with code the concepts explained in class | 🚨task1: answer what is moshe's desired knowledge state? | | 2. check-in task 🚨🚨goal2: let students infer their knowledge state | | building on outcome of 🚨task1, the transcript from the last recitation, design a task that students can digest 🧙‍♂️🍱 moshe's recipe of cuisines and implement with 💻🍱 angie's recipe for programming. before - 👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am - 👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain - 🏢city council's hypothesis: citizen's car ownership are less likely to support 🚸pedestrianization than non-car owners - 🌙amoon's hypothesis: 👨‍🎓student's 📚assignment load and 😋hungriness affect computer lab attendance | 🚨🚨task2: what are three prompts i should give to students to help students learn their knowledge state? | | 3. discuss to collect data 🚨🚨🚨goal3: angie learns students' knowledge state | 🤩angie's inference and allocation 1. desire: infer moshe's desired knowledge state of students (review class), infer students current knowledge state (check-in+data collection), prepare students to bridge that gap by programming case study () 2.output measure: by your grades (customer: moshe) and course evaluation (customer: students) 3. reward: quality of case study 1~5 submission | building on outcome of 🚨task1, 🚨🚨task2, 🤩angie's inference and allocation, design one data i should collect, given my goal to maximize average delta K (knowledge gain) from student participating the recitation. last class, I collected self-reported participation. | 🚨🚨🚨task3: what should angie focus on when listening to student's answer to learn their knowledge state? | | 4. preview case study 🚨🚨🚨🚨goal4: angie helps students reach to moshe's desire from each of your knowledge state | [[CS1_25.pdf]] | building on outcome of 🚨task1, 🚨🚨task2, 🚨🚨🚨task3, explain in three sentence how i should curate case study to students to motivate them. | 🚨🚨🚨🚨task4: how can angie bridge students' knowledge state with moshe's desire with programming in case study? | ## r4 made [claude project](https://claude.ai/project/84f30295-177b-4175-a886-49d7eaf4971b) # ⏰time allocation | Planning Component | One-Sentence Instruction | Example from Your Framework | Time Allocation | | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | 1. Review Class | Synthesize lecture transcripts and notes into a structured framework presenting three key theoretical concepts with visual representations (Moshe's recipe), three instructor insights (Moshe's wisdom), and corresponding programming implementations (Angie's recipe). | "Project information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to two axes: 🧙‍♂️🍱 moshe's recipe of cuisines (three key ideas with diagrams wrapped in three moshe's wisdom) and 💻🍱 angie's recipe for programming (code visualizations of concepts)." | **Prep:** 2 hours **Class:** 15 minutes | | 2. Check-in Task | Design a hypothesis-driven real-world problem that requires students to apply the week's key concepts through programming implementation, following the pattern of your examples. | "Design a task like '👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am' or '👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain' that students can implement using programming." | **Prep:** 2 hours **Class:** 15 minutes | | 3. Data Collection | Design and implement an exclusive in-person data collection experience that provides immediate value to attendees while measuring specific knowledge gains beyond self-reported participation. | "Design one data collection method to maximize average delta K (knowledge gain) from students participating in the recitation, improving upon previously collected self-reported participation." | **Prep:** 4 hour - grading to give feedback (⭐️will prioritize who attend recitation) **Class:** 10 minutes | | 4. Preview Case Study | Create three concise sentences that explicitly connect the theoretical concepts from lectures, the practical skills from the check-in task, and the relevant applications in the course case study. | "Building on review class and check-in task, explain in three sentences how case study connects with the outcomes of previous components." | **Prep:** 2 hours **Class:** 15 minutes | | | **TOTAL** | | **Prep:** 10 hours **Class:** 55 minutes | # 🤜input2algorithm2output | each section's desire | input | algorithm | output | | ----------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- | | 1. review class 🚨goal1: infer moshe's desired knowledge state together | 🗣️ transcript1 and 2 [[1.202 - Demand Modeling Feb24_otter_ai.txt]] [[1.202 - Demand Modeling Feb26_otter_ai.txt]] 📝lecture note 1,2,3 [[1.202_s25_Lecture6_LinearRegression.pdf]] [[1.202_s25_Lecture7_MultivariateRegression.pdf]] [[1.202_s25_Lecture8_GoodnessofFit.pdf]] | project 🗣️ transcript1,2 that explains 📝lecture note 1,2,3 to my three 🤜action space in the following three steps 1. imagine an information space spanned by 🗣️ transcript1,2 that explains 📝lecture note 1,2,3. 2. imagine two axes: 🧙‍♂️🍱 moshe's recipe of cuisines, 💻🍱 angie's recipe for programming. 3. project the information space from 1 to the axes in 2. outcome of 3 should provide the most efficient review of what moshe (instructor) intended students to learn about demand modeling in a format of teaching programming. 🧙‍♂️🍱 moshe's recipe of cuisines consists of💡three key ideas (with higher preference on the concepts with diagram from 📝lecture note) wrapped in a 🧙‍♂️ three moshe's wisdom angie found interesting (moshe wants to teach us to be a chef, not instill recipe) 💻🍱 angie's recipe for programming shows (visualizes) with code the concepts explained in class | 🚨task1: answer what is moshe's desired knowledge state? | | 2. check-in task 🚨🚨goal2: let students infer their knowledge state | [[TA 1.202 - R3 Demand Modeling_otter_ai.txt]] | building on outcome of 🚨task1, the transcript from the last recitation, design a task that students can digest 🧙‍♂️🍱 moshe's recipe of cuisines and implement with 💻🍱 angie's recipe for programming. before - 👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am - 👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain - 🏢city council's hypothesis: citizen's car ownership are less likely to support 🚸pedestrianization than non-car owners - 🌙amoon's hypothesis: 👨‍🎓student's 📚assignment load and 😋hungriness affect computer lab attendance | 🚨🚨task2: what are three prompts i should give to students to help students learn their knowledge state? | | 3. discuss to collect data 🚨🚨🚨goal3: angie learns students' knowledge state | 🤩angie's inference and allocation 1. desire: infer moshe's desired knowledge state of students (review class), infer students current knowledge state (check-in+data collection), prepare students to bridge that gap by programming case study () 2.output measure: by your grades (customer: moshe) and course evaluation (customer: students) 3. reward: quality of case study 1~5 submission | building on outcome of 🚨task1, 🚨🚨task2, 🤩angie's inference and allocation, design one data i should collect, given my goal to maximize average delta K (knowledge gain) from student participating the recitation. last class, I collected self-reported participation. | 🚨🚨🚨task3: what should angie focus on when listening to student's answer to learn their knowledge state? | | 4. preview case study 🚨🚨🚨🚨goal4: angie helps students reach to moshe's desire from each of your knowledge state | [[CS1_25.pdf]] | building on outcome of 🚨task1, 🚨🚨task2, 🚨🚨🚨task3, explain in three sentence how i should curate case study to students to motivate them. | 🚨🚨🚨🚨task4: how can angie bridge students' knowledge state with moshe's desire with programming in case study? | --- hypothesis - student's hungriness (X1) and number of assignments due (X2) affect recitation participation (Y) P(H0) my incentive is tilted toward money: number of assignment: - hypothesis-driven approach to model specification - input: important questions from last week - how is hypothesis testing relevant to our decision making? hypothesis driven approach to model specification (identifiability and sanity checks) [[theory-based models.png]] your output format (with my draft) two-sided vs one-sided test: if i give snack or lunch, hungriness ## r3-hypothesis testing sampling # input | | prompt for plan | documents | | --------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | 1. review class | based on the two attached transcript, extract the following: - 💡three key ideas (with higher preference on the concepts with diagram from 📝lecture note) - 🧙‍♂️ three moshe's wisdom angie found interesting (moshe wants to teach us to be a chef, not instill recipe) give plan1 on how to review class's | 🗣️ transcripts [[1.200 hypothesis testing_otter_ai.txt]], for 📝lecture note [[1.202_s25_Lecture4_Hypothesis (2).pdf]] 🗣️ transcripts [[1.202 sampling.txt]] for 📝lecture note [[1.202_s25_Lecture5_Sampling.pdf]] | | 2. check-in task | make one example with four hypothesis, in the best form so that students can understand the 💡three key ideas and 🧙‍♂️three moshe's wisdom from 1. re-review class - 👩🏽‍🍳chef's hypothesis: later shift 👨🏻‍🚒workers are more likely to order 🍔lunch items at 10am - 👨🏼‍💼marketer's hypothesis: umbrella usage of 👨‍👩‍👧‍👦 boston local's during snow is lower than during ☔️ rain - 🏢city council's hypothesis: citizen's car ownership are less likely to support 🚸pedestrianization than non-car owners - 🌙amoon's hypothesis: 👨‍🎓student's 📚assignment load and 😋hungriness affect computer lab attendance | | | 3. preview case study | summarize case study attached and give plan2 on how to cover this in two sessions - mle | [[CS1_25.pdf]] | | 4. data collection | given my goal is to make students participate in recitation, what data should i collect from students? | | # output | | plan | | -------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | check-in task's plan1 | 💡five key ideas 1. hypothesis testing: 2. mle for five key distributions: 🧙‍♂️three moshe's wisdom to make us a chef, not instill recipe 3. error rate = alpha * p(H0) + beta * p(H1); .18 | | preview case study's plan2 | summarize case study1 attached and give plan2 on how to cover this in two sessions | | re-review class's plan3 | based on the two attached transcript, give plan3 on how how to | | data collection | e.g. given your choice of snack type (0,1,2 for cranberry biscuit, cranberry ball, cookie), how many assignments is due this week? | 2025-02-21 | Section | Key Tasks | Angie's investment | | ------------------------ | --------------------------------------------------------------------------------------- | ------------------------------------ | | check in to review class | - Explain participation metrics and hypothesis - Link participation to success rates | 2 hr | | preview case study | Demo participation tracking and probability calculations | 3 hrs: find chellange you might face | | re-review class | | 0 hr | | Planning for next week | Set up tracking for next session | 5 hrs: Review data and create plan | | 🟦Blue I. Statistics intro | 🟩Green II. Linear regression models | 🔶Orange III. Fundamentals of Discrete Choice Models V. Discrete Choice Models (cont.) VI. Logit Mixture and Stated Preferences Data | 🔴Red VII. Machine Learning | | --------------------------------------------------------------------------------- | --------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------- | | 1. Introduction to Demand Modeling | 6. Least Squares Regression and Its Properties | 13. Introduction, Choice Behavior, and Discrete Choice Models | 15. Aggregate Forecasting, Microsimulation, Iterative Proportional Fitting | | 2. Probability Review | 🌙R3. Hypothesis Testing 📊Case Study 1 | 14. Specification and Estimation of Logit Models | 16. Statistical Tests of Model Specification | | 🌙R1. Probability Review 📊Case Study 0 | 7. Multivariate Regression | 🌙R7. Introduction to BIOGEME; 📊Case Study 3 | 🌙R8. Logit Estimation and Testing; Aggregate Forecasting | | 3. Statistical Inference; Maximum Likelihood Estimation; Properties of Estimators | 8. Goodness of Fit & Statistical Tests | | 17. Model Estimation Under Alternative Sampling Strategies, Endogeneity | | 4. Hypothesis Testing | 🌙R4. Linear Algebra Review; Linear Regression | 18. Nested Logit | | | 🌙R2. Maximum Likelihood Estimation; Properties of Estimators | 9. Violations of OLS Assumptions | 🌙R9. Nonlinear Specifications; Specification Testing 📊Case Study 4 | 25. Machine Learning for Regression Models | | 5. Sampling | 10. Generalized Least Squares | 19. MEV Models | 🌙R13. Exam Review | | | 🌙R5. OLS & GLS 📊Case Study 2 | | 26. Machine Learning with Theoretical Constraints | | | 11. Time Series Analysis | 20. Mixture Models | | | | 12. Use of Instrumental Variables, Simultaneous Equation Models | 🌙R10. MEV Models | | | | 🌙R6. Midterm Exam Review | 21. Simulation-based Estimation | | | | | 🌙R11. Mixture Models 📊Case Study 5 | | | | | 22. Bayesian Estimation and Discrete Choice | | | | | 23. Stated Preferences Methods | | | | | 🌙R12. Bayesian Estimation; Combining Data | | | | | 24. Practical Issues in SP Design and Combining Revealed and Stated Preferences | |