Forecast sales with Entity Embedding LSTM
Project description
AM18_SPR20_LondonLAB
Package Description
The package contains one object Kami with four methods in the order of execution:
- Preprocess()
- Analyse(n_sample)
- Vis()
- Forecast(store_list, product_list, start, end)
To initiate the object Kami, the user is required to supply at least these three arguments:
- Path to the grouped product sales input data (input_f_path)
- Path to an intermediary folder to store intermediary data (cache_dir_path)
- Path to an output folder to store final predictions (output_dir_path)
While Preprocess and Vis methods are executed without any argument, Analyse method can be supplied with an optional argument n_sample which is the number of random samples drawn from the predefined training data.
Forecast method is required to be supplied with four arguments including:
- A list of stores whose sales are predicted (store_list)
- A list of products whose sales are predicted (product_list)
- The start date of the forecast (start)
- The end date of the forecast (end)
Typical Use Case
from Kami import Kami
obj = Kami(input_f_path = 'PATH_TO_SALES_DATA/SALES_DATA.csv',
output_dir_path = 'OUTPUT_FOLDER/',
cache_dir_path = 'CACHE_FOLDER/')
obj.Preprocess()
obj.Analyse()
obj.Vis()
obj.Forecast(store_list = ['STORE_A', 'STORE_B'],
product_list = ['PRODUCT_A', 'PRODUCT_B'],
start = 'MM/DD/YYYY',
end = 'MM/DD/YYYY')
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Kami-0.4.3.tar.gz
(9.9 kB
view hashes)