3

Some feedback/tips/tricks/opinions here: Problem setup Including requirement analysis. Gotta decide how the system/solution should work, how to know ho how well we are doing, and then how to get there. Model evaluation. It is very desirable to have a quantitative way to evaluate our model performance. For that we want some labeled data. It is very quick to ...


1

ARIMA should work fine - I'd recommend the auto-arima package here: https://pypi.org/project/pmdarima/ Alternatively, if you're happy with it to be a little more opaque, you could use Facebook's Prophet. Generates some really accurate predictions with very little tuning.


1

ARIMA could work, I think it's the right approach. It's simple enough to be used on a small dataset, but sufficiently flexible at the same time. If you are using Python, library statsmodels allows you to implement ARIMA regressions. You have to grid search and find the right parameters to find the best fit, and run the prediction. If you want to know how to ...


1

You can use a total variation regularizer (https://en.wikipedia.org/wiki/Total_variation_denoising), it's a penalty for abrupt changes of neighbor values. It's usually used for images, that's why its TF version (https://www.tensorflow.org/api_docs/python/tf/image/total_variation) operates with 4D tensors, but if you're writing your model in pytorch for ...


1

Alternatively, you can use pd.cut to create your desired bins and then count your observations grouped by the created bins. from faker import Faker from datetime import datetime as dt import pandas as pd # Create sample dataframe fake = Faker() n = 100 start = dt(2020, 1, 1, 7, 0, 0) end = dt(2020, 1, 1, 23, 0, 0) df = pd.DataFrame({"datetime": [...


Only top voted, non community-wiki answers of a minimum length are eligible