felgueres/predict-shopping-time

Predicting shopping time

Modeling Shopping Time

Date: July 28th, 2017

Description:

Modeling shopping time duration for inferece, includes:

1. understand the data via exploration (EDA),
2. design a workflow to transform raw data into the feature space of the model,
3. build model and predict
4. extra: inference of resulting feature space and model

Model:

• L1 Regularized Regression (Lasso)
• Random Forest (minor exploration)

Code:

preprocessing.py - contains Preprocessing class to:

• Preprocessing of data prior to model fit
• Returns feature engineered datasets
• Handles dummy variables without leakages
• Handles scaling

model.py

• Loads data
• Featurizes data
• Runs Lasso Model
• Outputs predictions into folder ./predictions/

File usage for predictions of the test-set:

• In console, navigate to ./src/ folder.
• run -model.py

Used Technologies:

• Python
• Pandas
• Matplotlib / Seaborn
• sklearn
• StatsModels
• Jupyter Notebook