search

data-describe / awesome-data-science-models

A few end to end examples that use data-describe
Apache License 2.0
16 stars 17 forks source link

readme

Awesome Data Science Models

The goal of these projects is to provide tutorials (and answers) of our most popular models we use for training.

They all use data⎰describe our open source accelerator for EDA.

Library Item Title Description Audience
101 Baseball Pitch Predictor Kubeflow pipeline for modeling around pitch type Citizen Data Scientists, Advanced Data Scientists
102 Beatles Like Predictor Predict who likes the Beatles using purely GCP AutoML Tables Citizen Data Scientists
201 Lending Club bad-loan predict if a loan applicant is high risk using XGBoost and hyper-parameter training. Data Scientists, Data Engineers
301 Census Income The census income example is a logistic regression model used to demonstrate Google Cloud Platform's AI Platform. Data Scientists, Data Engineers
202 Taxi Cab Prediction A wide and deep neural net implemented using Tensorflow predict trip duration for Chicago taxi rides. Data Scientists, Data Engineers
203 Black Friday Recommending product categories with multi-class categorization in AI Platform with custom prediction routines, custom hyperparameter tuning Data Scientists, Data Engineers
302 Cellular Imaging Recursion Cellular Image Classification with AI-Platform with TF2, TPU, and advanced Engineering Data Scientists, Data Engineers, HCLS
401 NASA IOT Signal processing Processing Streaming data with signal windowing and live prediction using AI Platform and GCP MLOps, Advanced Data Scientists
Bearing Condition Monitoring Analyzing the Vibrations and Motor current of Bearings to predict Remaining Useful LifeCycle (RUL) using Vertex AI Data Scientist, Data Engineers

install instructions

These examples are meant to run on GCP AI Platform. They may very well run elsewhere but we haven't tested.

Create an instance for AI Platform notebooks:

  1. Choose Tensorflow Enterprise 2.1 (No GPUs)
    • Make sure you have 4 CPUs and at least 15 GB of memory
    • Click Open JupyterLab
    • Use the Launcher (right-hand-side of screen) to open a Terminal...
  2. Install data describe:
    • pip install data-describe[all]
    • pip install xgboost==0.82
    • pip install pandas_gbq
    • pip install google-cloud-bigquery
    • pip install google-cloud-storage

Clone the examples:

git clone https://github.com/data-describe/awesome-data-science-models.git

Sources of Data

Beatles Like Predictor

The original dataset is available from listenbrainz and provided by BigQuery here

Black Friday

The original dataset is on Kaggle, and can be found here.

Census Income

The original dataset is on the UCI Machine Learning Repository, and can be found here.

Chicago Taxi

The original dataset is a BigQuery public dataset, and can be found here. More information on BigQuery public datasets can be found here.

Lending Club

The original dataset is made public by LendingClub, and can be found at their website or here. The dataset used for this demo is a subset of the original dataset.

Cellular Image

The original data is part of the tensorflow datasets. We have copied this data into a public bucket for the demo here.

Nasa IOT Data

J. Lee, H. Qiu, G. Yu, J. Lin, and Rexnord Technical Services (2007). IMS, University of Cincinnati. "Bearing Data Set", NASA Ames Prognostics Data Repository (http://ti.arc.nasa.gov/project/prognostic-data-repository), NASA Ames Research Center, Moffett Field, CA

Bearing Condition Monitoring

The original data is a public dataset of the Chair of Design and Drive Technology, Paderborn University. It is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License and it is available for downloading it here