Time-Series Forecasting

Load & explore Time-Series data:

(header=0: We must specify the header information at row 0, parse_dates=True: We give the function a hint that data in the first column contains dates that need to be parsed, index col=0: We hint that the first column contains the index information for the timeseries, squeeze=True: We hint that we only have one data column and that we are interested in a Series and not a DataFrame)

• How to load your time series dataset from a CSV file using Pandas.
• How to peek at the loaded data and query using date-times.
• How to calculate and review summary statistics.

Basic Feature Engineering:

• Date Time Features
• 
• Lag Features (shift())
• Rolling Window Statistics (rollling(window=2)
• Expanding Window Statistics

Data visualization:

• Line Plots (pd.grouper())
• Histograms and Density Plots.
• Box and Whisker Plots.
• Heat Maps.
• Lag Plots or Scatter Plots.
• Autocorrelation Plots.

Resampling & Interpolation:

resample('D') to resample & interpolate() function to interpolate missing values.

Power transforms:

• Identify a quadratic change and use the square root transform.
• Identify an exponential change and how to use the log transform.
• Use the Box-Cox transform to perform square root and log transforms andautomatically optimize the transform for a dataset.

Moving Average

• Simple Moving average
• Weighted Moving average
• Exponential moving average
• Explonetial smoothing moving average

Introduction to White Noise

A time series is white noise if the variables are independent and identically distributed with a mean of zero. This means that all variables have the same variance (sigma2) and each value has a zero correlation with all other values in the series.

• White noise time series is defined by a zero mean, constant variance, and zero correlation,
• If your time series is white noise, it cannot be predicted, and if your forecast residuals are not white noise, you may be able to improve your model.

Introduction to the Random Walk

• Random Walk and Autocorrelation.
• Random Walk and Stationarity.

Decompose Time Series Data:

• y(t) = Level + Trend + Seasonality + Noise, y(t) = Level * Trend * Seasonality * Noise

Use and Remove Trends

• The importance and types of trends that may exist in time series and how to identify them.
• How to use a simple differencing method to remove a trend.
• How to model a linear trend and remove it from a sales time series dataset.

Use and Remove Seasonality

Stationarity in Time Series Data

Backtest Forecast Models

Forecasting Performance Measures

Persistence Model for Forecasting

Visualize Residual Forecast Errors

Reframe Time Series Forecasting Problems

Introduction to the Box-Jenkins Method

Autoregression Models for Forecasting

Moving Average Models for Forecasting

ARIMA Model for Forecasting

Autocorrelation and Partial Autocorrelation

Grid Search ARIMA Model Hyperparameters

Forecast Confidence Intervals

Prepare "Time Series" data for CNNs and LSTMs:

Develop CNNs for Time SeriesForecasting: