| """Script that generates, pickles and stores posterior data""" | |
| import pickle | |
| import numpy as np | |
| import config as cfg | |
| from scripts.posterior import PosteriorGenerator | |
| np.random.seed(42) | |
| if __name__ == '__main__': | |
| demo_prices = np.concatenate([np.repeat(10,10), np.repeat(7.5,25), np.repeat(11,15)]) | |
| possible_prices = np.linspace(0,20,100) | |
| for el in [x/100 for x in range(5,100,5)]: | |
| demo_demands = np.exp( | |
| np.random.normal( | |
| loc=-el*demo_prices+cfg.LATENT_SHAPE, | |
| scale=cfg.LATENT_STDEV, | |
| ) | |
| ) | |
| ts = PosteriorGenerator(prices=demo_prices, demands=demo_demands) | |
| posterior = ts.calc_posterior(samples=5000) | |
| post_demand_samples = [] | |
| for idx in range(len(posterior)): | |
| elas = posterior.get_values("elas")[idx] | |
| shape = posterior.get_values("shape")[idx] | |
| post_demand_sample = np.exp(elas*possible_prices + shape) | |
| post_demand_samples.append(post_demand_sample) | |
| with open(f"assets/precalc_results/posterior_{el}.pkl", 'wb') as f: | |
| pickle.dump(post_demand_samples, f) | |