/BloodGlucoePredictor

Primary LanguagePython

BloodGlucoePredictor

In this project, we have to predict the Blood Glucose level for an hour in the future at every 5 minutes interval.

-- Project Status: [Completed]

Project Intro/Objective

The purpose of this project is to predict the Blood Glucose level.

Methods Used

  • Machine Learning
  • Data Visualization
  • Predictive Modeling
  • Deep Learning
  • Auto Regressor, ARIMA & LSTM

Technologies

  • Python
  • Pandas, jupyter, Numpy, Keras, Tensorflow

Project Description

I have tackled this problem from a Time Series perspective and tried different machine learning methods. At the very first glance,I have tried Auto-regressor & ARIMA models but after digging deeply inside the available datasets and realize the shortage of feature set and week correlations b/w different features I decided to use Deep Learning and specifically choose LSTM, which gave me the state-of-the-art results for this model.

Getting Started

  1. Clone this repo (for help see this tutorial).

  2. Raw Data is being kept [/data](Repo folder containing raw data) within this repo.

  3. Data processing/transformation script is being kept [/src/data/preprocess.py]

  4. Model implementation functions are kept at [src/models/*]

  5. Clark Error Grid implementation can found at [src/visualization]

  6. The trained and saved model can be found at [/models]

  7. Some jupyter notebooks for analysis can be found at [/notebooks/]

  8. Plot images can be found at [/plot_images/] 9 All requirements are added to the [requirements.txt] file.

Contributing DSWG Members

Team Leads (Contacts) : AbdulRehman(@Abdul)

Other Members:

Name Slack Handle
Abdul Rehman @Abdul

Important Note:

I have achieved the required result(accuracy with RMSE & Clark Error Grid), but I believe I can improve this model a lot if I got enough time.

Contact