codesigma91 / project_machinelearning

Machine Learning Project based on a housing dataset from Ames, IA for a Kaggle.com competition

Project_MachineLearning

Advanced Regression Techniques on Housing in Ames, IA

This project revolves around the kaggle.com competition on Advanced Regression Techniques, found here. The project goal was to develop and test regression models designed to predict home sale prices, based on a large number of descriptive features.

• A brief blogpost can be found here
• and our group's presentation can be found here.

This Python machine learning project is the work of the "Bench Initiative" group from the winter 2019 cohort of the NYC Data Science Academy. Our members are:

• Eric Adlard
• Ryan Essner &
• Sabbir Mohammed

This repository holds our Python 3 Jupyter notebooks and relevant files. Here is a brief workflow explaining the file and folder structure:

• A) The "house-prices-advanced-regression-techniques" folder initially contains the raw train and test data sets from kaggle.com. Inside it are a few additional files(.csv) used by our group to organize our work.

• B) The "1. Training set overview" Jupyter notebook shows our brief, initial exploration of the training data.

• C) The "2. Imputation and Feature Engineering" notebook holds our code where we handled missingness (in both the test and train sets) and how we updated the features to be analyzed. It outputs the .csv(s) used for modeling, into the "data" folder.

• D) The "3. Feature Reduced data set" notebook holds similar code to the imputation notebook, but instead it outputs .csv files of the reduced data sets into the "data" folder as well.

• E) Finally, the remaining notebooks (#4 and #5) carry out regression modeling done on the final, prepared data sets from the "data" folder, and outputs formatted csv(s) for submission into the kaggle.com competition to the "kaggle_submissions" folder.

---

Thank you for visiting our repository!