garethwilhelm / ts-feature-clustering Goto Github PK

Project has 3 major phases for 2 major experiments (i. export value and ii. rca values):

1. Exploratory data analysis: to check overall trends, patterns, outliers and suspicious data

   • 

   • ![Histograms: to illustrate frequency distributions] ()

2. Time series characterisation: to check whether or not time series have trends, are stationary, are seasonal, are cyclical or random to inform the decision on appropriate models

   • Autocorrelation and partial autocorrelation plots: Check existence of trends, cyclical, stationarity and/or seasonality patterns ()

   • Moving average

   • Augmented Dickey-Fuller Test

3. Time series clustering and classification In progress: Literature review:

   • Feature extraction
   • Feature selection
   • Clustering model selection
   • Clustering

4. Time series forecasting

• Planning: