movie_selection
A project organized around understanding decision making and movie selection.
Directory Tree
.
.
├── ddm_analysis
│ ├── data
│ │ ├── study1.csv
│ │ ├── study1_measured_mood.csv
│ │ ├── study2.csv
│ │ └── study3.csv
│ ├── power_simulate_HDDM.ipynb
│ ├── step1_data_pre-processing
│ │ ├── stimulus_check.R
│ │ ├── study1_pre-processing.R
│ │ ├── study2_induction_check.R
│ │ └── study2_pre-processing.R
│ ├── step2_Fitting_HDDM_study_1.ipynb
│ ├── step3_Inference_testing_study_1.ipynb
│ ├── step4_Fitting_regression_study_2.ipynb
│ ├── step5_Inference_testing_study_2.ipynb
│ ├── step6_Cell_means_study_2.ipynb
│ ├── step7_convergence_test
│ │ ├── convergence_study1.ipynb
│ │ └── convergence_study2.ipynb
│ ├── step8_plot
│ │ ├── plots.ipynb
│ │ ├── study1_correlation_traces.csv
│ │ ├── study_1_ddm_confirmative_analysis.csv
│ │ ├── study_1_ddm_exploratory_analysis.csv
│ │ ├── study2_cell_mean_trace_df.csv
│ │ └── study2_full_regression_hypothesis.csv
│ └── study3_analysis
│ ├── s3_analysis.ipynb
│ ├── s3_data_preprocessing.R
│ └── study3.csv
├── decision_task_pavlovia
│ ├── study_1
│ │ ├── index.html
│ │ ├── media_decision_making.js
│ │ ├── README.md
│ │ └── resources
│ │ └── SAM
│ └── study_2
│ ├── index.html
│ ├── media_decision_making_group_1.js
│ ├── README.md
│ └── resources
│ └── README.m
├── generate_summaries
│ ├── anew
│ │ ├── anew_sentiment_analysis.py
│ │ ├── english_shortened.csv
│ │ └── Output Anew Sentiment movie_summaries.csv
│ ├── generate_summaries.ipynb
│ ├── MovieSummaries
│ │ ├── character.metadata.tsv
│ │ ├── movie.metadata.tsv
│ │ ├── name.clusters.txt
│ │ ├── plot_summaries.txt
│ │ ├── README.txt
│ │ └── tvtropes.clusters.txt
│ ├── movie_summaries_scored_and_selected.ods
│ ├── movie_summaries.txt
│ └── plot_summaries.txt
├── LICENSE
├── README.md
└── summary_scoring
├── combo.csv
├── raw_text.csv
└── reading_ease.ipynb
Generate Movie Plot Summaries
This code takes a corpus of movie summaries and uses NLP to reduce these summaries into several short sentences. You can think of these as summaries of movie summaries. Once summaries are generated, the ANEW dictionary is used to characterize the arousal, dominance,and valence score for each summary.
Resources used:
Summary Scoring
Code to analyze the read easiness of the generated movie summaries.
Decision Task Pavlovia
The code to replicate the movie decision making task on Pavlovia. This experiment is implemented by PsychJS, and hosted on Pavlovia.
DDM analysis
The code to replicate the analysis are organized in sequential steps.
Data directory
Data csv files are pre-processed data, which is the output of step-1.
subj_idx: index for each subject stim: decision type rt: reaction time response: 1 if higher boundary choice selected, 0 if lower boundary choice selected mood_valence_measured: measured mood valence for each participant in each block mood_arousal_measured: measured mood arousal for each participant in each block
Step 1 Data pre-processing
R
Input file is raw data files collected from Qualtrics and Pavlovia. The code pre-processes the raw data into csv files for study 1 and study 2. It also includes an R-script for the induction check & stimulus check, including the plot and ANOVA testing.
Step 2 (study 1) & 4 (study 2) Model fitting
Jupyter Notebook
Loads the pre-processed data csv files from step 1. And fitts the HDDM models, then outputs saved models for inference testing.
Step 3 (study 1) & 5 (study 2) Inference testing
Jupyter Notebook
Loads the fitted model from step 2 & 4, and output the statistics of the posteriors distributions, including inferences for hypothesis testing, and the 95% credible interval.
Step 6 (study 2) Generate samples for plotting
Jupyter Notebook
Loads the data for study 2, and fitted HDDM for each individual cell. It outputs drift rate MCMC samples for each cell, and saves the samples to csv file for plotting.
Step 7 Model convergence
Jupyter Notebook
It fits each model 4 times, and calculate inter-chain variability (GR metrics) to test the convergence of HDDM models.
Step 8 Plotting
Jupyter Notebook
The directory includes a notebook file which reproduce plots in the manuscripts. The data for plotting is also included in the directory.
Requirement
| R | Jupyter Notebook |
|---|---|
| ggplot2 | hddm |
| gridExtra | matplotlib |
| tidyverse | pandas |
| rstatix | numpy |
| hrbrthemes | kabuki |
| viridis | seaborn |
| ggpubr | scipy |