Please enable JavaScript.
Coggle requires JavaScript to display documents.
TS (ETL), Forecast/Predict # # # - Coggle Diagram
TS
ETL
Naive
Use Naive method such as
the previous value (shift 1) as prediction
error metric
Calculate error metrics
store error metrics with metadata
Split Data
Training Set
Convert to Tensor Dataset
Choose Feature Derivation Window Size
DNN Model Fit
Hyperparameters tuning:
Window Size
Batch Size
Learning Rate
Shuffle buffer
Epoch
error metric
LSTM Model Fit
error metric
RNN Model Fit
error metric
Test Set
FIT ARIMA model
#
error metric
Calculate error metrics
store error metrics with pdq metadata
Find ARIMA orders
#
Calculate BIC score for each combinations pdq
Find best pdq candidate (top 3)
Store pdq as metadata
Resampled (quarterly)
Impute Missing (ffil)
Subset (Duration years)
Forecast/Predict
#
#
#