Zero to Hero Machine Learning Pipeline: London Airbnb Open Data

Introduction

"He was no one, Zero, Zero Now he's a honcho, He's a hero, hero!" Isn't this familiar? Absolutely it is. No way you could forger about Disney's cool animation, Hercules.

Source

"Zero to Hero" is a phrase that is often used to describe a transformation from a novice or beginner to an expert or accomplished individual. It can also refer to a person who starts with very little and through hard work and determination, becomes successful. Throughout this journey, I want to take you to a step-by-step guide on building a regression Machine Learning pipeline.

This mini-project is about building a regression pipeline to predict the number of reviews per month of each Airbnb property, trained based on a dataset collected on September 2022 in London.

Dataset

The dataset to build this pipeline was captured from Detailed Airbnb Listing Data (London, Sep 2022) on Kaggle. The original data was prepared by Inside Airbnb project. The mission of Inside Airbnb is to empower residential communities with data and information that enables them to understand, make decisions and have control over the effects of Airbnb's presence in their neighborhoods.

Proposed ML pipeline

This regression ML pipeline is built for predicting number of reviews per month on London's Airbnb data. Predicting this number is important because it can help AirBnb to better understand and manage customer engagement. This can have a number of benefits, such as improving customer satisfaction, optimizing marketing efforts, increased revenue, and forecasting.

Sharing the Results

This pipeline is created in Python, using a group of ML models from scikit-learn library. For the sake of reproducibility, all of the pipeline is scripted and the results will be stored in relevant directories. There is a notebook to report the performance of the various models which uses the stored results along with presenting them through visualizations. In order to render all the plots properly, a HTML version of this report has been updated too. The report can be found here.

Usage

This pipeline can be reproduced by cloning the GitHub repository, installing the dependencies listed below and running the following commands at the terminal from the root directory of this project. To run this model follow the below steps:

Step 1: Clone this repository by running the code below in CLI:

git clone [email protected]:mrnabiz/zero-to-hero-ml-pipline.git

Step 2: Install the Conda environment by running the code below:

conda env create -f environment.yml

Now you have the required files and dependencies to reproduce the pipeline and deploy it. To build the pipeline and render the report, you have two options:

1. Run the commands, generate the results, and store them step-by-step (To understand the pipeline building stages, I highly recommend you choosing this option.)
2. Run the commands all in once with make command

Option 1: Running step-by-step scripts to build th pipeline

Navigate to the repository's root folder in CLI and follow the steps below:

Step 1: Download the raw data by running src/data_wrangling/pull_data.py:
--file_path should be the path where the data will be saved
--url should be the link to the data

python src/data_wrangling/pull_data.py --file_path="data/raw/raw_df.csv" --url="http://data.insideairbnb.com/united-kingdom/england/london/2022-09-10/data/listings.csv.gz"

Step 2: Clean the raw data frame to remove unnecessary columns by running src/data_wrangling/clean_data.py:
--output_file_path should be the path where the clean data will be saved
--input_file_path should be the path where the raw data is stored

python src/data_wrangling/clean_data.py --output_file_path="data/raw/clean_df.csv" --input_file_path="data/raw/raw_df.csv"

Step 3: Preprocess the cleaned to prepare it for the model building by running src/data_wrangling/preprocessing.py:
--output_file_path should be the path where the clean data will be saved
--input_file_path should be the path where the raw data is stored

python src/data_wrangling/preprocessing.py --output_file_path="data/preprocessed/preprocessed_df.csv" --input_file_path="data/raw/clean_df.csv"

Step 4: Run the EDA analysis by running src/model_building/eda.py:
--output_dir Path to folder where the EDA results will be saved, in quotes\ --clean_df_file_path Path to folder and file name where the clean df is stored, including the name,in quotes
--preprocessed_df_file_path Path to folder and file name where the preprocessed df is stored, including the name, `in quotes

python src/model_building/eda.py --output_dir="results/eda_plots" --clean_df_file_path="data/raw/clean_df.csv"  --preprocessed_df_file_path="data/preprocessed/preprocessed_df.csv"

Step 5: Run the feature engineering script by running src/model_building/feat_eng.py:
--output_file_path should be the path where the file with new features will be saved
--input_file_path should be the path where the preprocessed data is stored

python src/model_building/feat_eng.py --output_file_path="data/preprocessed/feat_eng_df.csv" --input_file_path="data/preprocessed/preprocessed_df.csv"

Step 6: Run the linear models by running src/model_building/linear_models.py:
--output_dir Path to folder where the model results dictionary will be saved, in quotes\ --input_file_path Path to folder and file name where the preprocessed file with engineered features is stored, including the name,in quotes

python src/model_building/linear_models.py --output_dir="results/model_results" --input_file_path="data/preprocessed/feat_eng_df.csv"

Step 7: Run the non linear models (ensembles) by running src/model_building/ensembles.py:
--output_dir Path to folder where the model results dictionary will be saved, in quotes\ --input_file_path Path to folder and file name where the preprocessed file with engineered features is stored, including the name,in quotes

python src/model_building/ensembles.py --output_dir="results/model_results" --input_file_path="data/preprocessed/feat_eng_df.csv"

Step 8: Run the permutation feature importances analysis by running src/model_building/perm_feat_importances.py:
--output_dir Path to folder where the model results dictionary will be saved, in quotes\ --input_file_path Path to folder and file name where the preprocessed file with engineered features is stored, including the name,in quotes

python src/model_building/perm_feat_importances.py --output_dir="results/model_results" --input_file_path="data/preprocessed/feat_eng_df.csv"

Step 9: Run the SHAP analysis by running src/model_building/shap_feat_importances.py:
--output_dir Path to folder where the model results dictionary will be saved, in quotes.\ --input_file_path Path to folder and file name where the preprocessed file with engineered features is stored, including the name,in quotes.

python src/model_building/shap_feat_importances.py --output_dir="results/model_results" --input_file_path="data/preprocessed/feat_eng_df.csv"

Option 2: Running the make file

Before running the pipeline through the make file, please use the code below to reset the repository-generated files. In the terminal, browse to the root folder of the repository and run:

make clean

After using the make clean command to reproduce the model pipeline, use the command below in the repository's root folder. Please ensure that you have followed the dependencies installation.

make all

Dependecies

I have created an environment (environment.yml) for Python 3.10.6 in which the reproduction of pipeline can be done. To create this environment on your machine run:

conda env create -f environment.yml

Activate it by running:

conda activate zero-to-hero-ml

License

The materials of this project are licensed under the MIT License. If re-using/re-mixing please provide attribution and link to this webpage.

References

Scikit-learn: Machine Learning in Python, Pedregosa et al., JMLR 12, pp. 2825-2830, 2011.