New answers tagged

1

A rate is always a gain per some time/step. A rate can exist even if the maximum is never reached. In supervised learning a loss function is defined, which is expected to have a global maximum, that we try to reach by gradient descent. How much closer we get with each timestep/iteration/epoch/batch is the rate of convergence.


0

Yes, you can predict who will open a loan with this model. You have to keep several things in mind. The model expects the new data to be from the same distribution as the training data, meaning the variables have the same distribution and targets as well. You can not train on one set where 3% opened a loan and then predict on a set where you expect that 90% ...


2

Generally, I'd pick a very simple, transparent/explainable model and use the results in a semi-automated way. That is, do not just derive a prediction but rather insights. You could, for example, use a (or multiple) decision tree(s) which you pre or post prune. The result could be a tree with, let's say, just 1-3 features to find simple rules like "if a ...


1

The main problem with very little data is that it's almost impossible to know how representative the sample is. Some people would even say that less 20-30 data points cannot be representative of anything. Every single data point can have a huge impact on any model, so any prediction has a huge margin of error. If one is going to train a model from a tiny ...


0

Given the domain is a digital card game, deck value and game play strategy can be simulated. The wide space of options can filled in with synthetic data. Classic machine learning algorithms, such as logistic regression and boosted trees, will have limited success given the sequential nature of the problem. It would be more useful to frame it as reinforcement ...


1

I suppose that you want to fit a logistic function to your data. A general form of logistic function is : $$y(x)=a+\frac{b}{1+c\: e^{-p\:x}}$$ So they are four parameters $a,c,b,p$ to optimize. The usual method is a non-linear regression calculus. This is an iterative process which requires 'guessed' initial values for the parameters to start the iteration. ...


1

They give you the current values of the model parameters $a$ and $b$, and a new data point $(x,y)$, and they request to perform one training step with gradient descent using the data point, and returning the updated values for $a$ and $b$. The problem with your code is that the sign of the learning rate is wrong in the parameter update. If you change it to ...


1

Firstly, when you have an imbalanced dataset accuracy is not a good metric to be using (see https://en.wikipedia.org/wiki/Precision_and_recall#Imbalanced_data). You should consider what the ultimate use-case of this model is and what metric is properly capturing the performance of the model considering that use case. For example, when classifying the ...


0

The goal is to identify at least 98% of the customers, that do not repay their debt. So the bank can "accept a maximum number of 'good' customers, that can be granted loans" Here the goal is focused on the bad customers. There should be at least 85% good customers accepted while the side focus is to reject as many bad customers as possible. I ...


Top 50 recent answers are included