New answers tagged

0

Modeling changes over time is often called dynamically modeling. There are many options to approach dynamical systems: Ignore it. Assume a static system and remove any data that changes over time. Model the time sensitive features. Encode the changes. For example create features like: income 1 quarter back, income 2 quarters back, … Use a time-based ...


0

I don't think you need a machine learning model for this. There's no data as such, which the machine can learn and give better predictions. If the user needs 70 % healthy meals in 14 days, we simply need to maintain a count of the healthy meals and the total meals consumed by the user. As we need 70 % healthy meals, the user must be prompted to have a '...


0

In the statsmodels, Omnibus is a test of the skewness and kurtosis of the residuals. The Prob(Omnibus) value ideally would close to zero which would indicate normalcy of the residuals. The Omnibus and Prob(Omnibus) are not a useful measures of how good a model is. One option to improve model fit is increase the number of features. Examples would include ...


0

If I am getting you right, you are trying to predict the absolute price, and not just the change in price from the last time step. This task is much more difficult than just predicting the change in price from the previous time step, as the net accurately needs to memorize what the price in the previous time steps were. I suggest you calculate the log ...


0

Let me start of by saying that the project sounds really cool. But here's the but: it is tremendously ambitious to accomplish what you want with the data you mention. Here's some stuff I think you would really need to consider, I'm sorry for the long post, but maybe it gets your creative juices flowing! What exactly do you want to know? You say you want to ...


0

If you would already have those datapoints before a company actually goes into bankruptcy then you can then them in your model since when predicting to the future you could have access to that data. However, if you would only know the data once the bankruptcy event happens (e.g. date of bankruptcy) then you cannot use this variable in your model since you ...


1

I assume the "dose" $y$ is limited to $y \in [0,1]$. So in the moment you have "bunching" in your target value $y$ which you try to remove. In this case, a linear regression could lead to "overshooting" (see here for more details). So it could be beneficial to use some estimator which "restrics" $\hat{y} \in [0,1]$ as ...


Top 50 recent answers are included