hpe-cty-app

Predicting cloud storage consumption using time-series analysis

• Application uses time-series based data for prediction of future trend of storage consumption

• Forecast trends for the specified time interval in the future on the trained model

• Storage consumption prediction in real-time : Model trains in real time while giving future prediction

• Built a web application using React as frontend and Flask as backend.

• Generated a dataset using python scripts for different usage patterns.

• Implemented a variety of forecasting models like ARIMA, Au- toArima, LR, XGB, RNN-LSTM and Prophet.

• Analyzed the ML models based on their MAE, MSE, R2 and RMSE scores. • Got an accuracy of around 85% on increasing-decreasing data and 95% on seasonal data.

Client Libraries

• ReactJS
• Axios
• React-google-charts
• Material UI

Backend(Python) Libraries

• Flask_Cors==3.0.10
• flask_restful==0.3.9
• matplotlib==3.5.1
• numpy==1.21.2
• pandas==1.3.2
• pmdarima==1.8.5
• scikit_learn==1.1.1
• tensorflow==2.9.1
• xgboost==1.6.1

Steps to run

Client

cd client
npm install
npm start

Flask backend

Create new environment(preferable, For conda: conda create --name myenv, conda activate myenv)
pip install -r requirements.txt
flask run