A project for predicting air raid alarms (everything could be done by performing each script step by step, how it's written in here(files with * can be skipped))
To run creation of the model run script "FullModelTraining"
To run creation of the current alarms run script "makingFileEachHour"
To run creation of the endpoint run "SaaS_alarms"
File for downloading data from the ISW site from 24.02.22 to 25.01.23
First part of preprocessing. Selecting main body from the html pages of ISW
Main part of preprocessing. Removing useless info and making text out of html
Vectorizing data received from the previous part of preprocessing for further fitting ML model
Script for getting weather for the selected place. Can be used with date to get info for 24 hours and without date to get info about next 12 hours
Script for creating dataframe with all words and adding weights of that words for each day(correlation matrix was also calculated here to make some EDA)
Merging all data from previous sources(weather, alarms, words, cities) to one dataframe
Splitting merged data into train and test data(also converting some string and datetime typed into floats)
Editing data so it would be possible to fit the model, training and testing model
Script for getting number of alarms of the day and number of alarms in the regions on specific time
Script for returning selected page
All scripts above for model training were added here to make full execution easier
Getting info about specific date and converting it into the dataframe with previous words
Collecting forecast for the specific date for all regions and editing it for further fitting into the model
Merging ISWPrediction and CollectForecast data
Script for predicting alarms for next 12 hours and saving results in file
Endpoint for getting info out of file with all alarms