This LSTM recurrent neural network learns to predict Croatia economic recessions via binary classification with various economic variables fed as inputs by training a model on past macroeconomic data and the current situation caused by Covid-19. All data used to build the predictor is taken from Croatian National Bank and Eurostat.
Data used to train the model begins in 1996 and ends in March 2020. I used large sets of data as inputs, like the unemployment rate by year, general government sector debt, GDP growth, most affected sectors by Covid-19 in terms of Government revenue.
What I ended up is a very complex algorithm that calculates how much will Croatian GDP fall (negative growth) and total government revenue in HRK (Croatian kuna).
My algorithm predicts that Croatian GDP will fall by –10.4% in 2020 and 2021 by –6.7%.
Later I will post all code so stay tuned.
I built a simple Nodejs watcher that is syncing changes...
const dataCsv = './sample_data.csv'; const mongodb = require("mongodb").MongoClient; const csvtojson = require("csvtojson"); const fs = require('fs'); let url = "mongodb://yoururl";
If you have to execute a function only if condition...
// You can use short-circuit condition && doSomething();
Designed & Built by Mijo Kristo