Time Series Forcasting

Time Series Analysis

• Data is spanned with "Time" variable

• Observations taken as specific time interval Day , Week , Month etc.

• Usages

  • Business Forecasting

  • Understand Past behaviors

  • Plan for future using forecast

  • Evaluate current accomplishments

Stationarity

A stationary time series is one whose properties mean, variance, autocorrelation, etc. do not depend on the time at which the series is observed. Time series with trends, or with seasonality, are not stationary

• Seasonality

  • Repeating pattern over fix period of time (Sales in festive seasons, Ice cream sales)

• Trends

  • Higher or lower movement over period of time

• Irregularity/ Residuals

  • Limited time random observation (Natural Disaster / Epidemics Medicine sales)

• Cyclic - Repeating pattern over random timeframes

Checking the Stationarity of Time Series

Moving average or moving variance

ADCF Test

• Null hypothesis is Time series in not stationary

• Calculate Test Statistics and Critical Values

Differencing :

Compute the differences between consecutive observations

ARIMA Model

Auto Regression Model (P)

• Correlation between previous and current time period (t-1 and t)

Integration (d)

Moving Average (Q)

Referances :

• https://towardsdatascience.com/time-series-of-price-anomaly-detection-13586cd5ff46

• https://towardsdatascience.com/anomaly-detection-for-dummies-15f148e559c1