Time-Series Predictive Analytics for Hair Device

• Aim of the project:
Stabilise the device temperature over a period of time based on the historical data to provide a ideal styling conditions to the user.

• Accomplishments:
– Carried out the data extraction from the device and performed data analysis on it.
– Optimized the parameter values for the ARIMA (non-seasonal),Auto-ARIMA, SARIMAX (seasonal) and LSTM model for better accuracy.
– Able to forecast for a very high frequency data (millisecond’s), which is impressive.

• Deliverable:
Stable temperature is ideal for hair styling, this framework successfully predicts the future power output (correlated with the temperature) to reduce the time lag.

• Breif results:
1. ARIMA:






2. Auto-ARIMA and SARIMAX:






3. LSTM: