Mario
- Member for 6 years, 8 months
- Last seen this week
- Berlin, Germany
revised
Overfitting detection
Keep the codes\arguments inside of the backticks for better understanding
Loading…
comment
Can LSTM be used for non time series data?
I have a problem with the mathematical logic you addressed in your answer. From my understanding, sequential time data means data is collected/recorded in equal time steps or lags. I still don't understand mathematically if the variable's coefficient $\beta$ within the regression formula was significant (means $p-value < 0.05$); thus, the data is sequential! How so? Do you have a reference for this logic?
revised
Can LSTM be used for non time series data?
Keep the codes\arguments inside of the backticks for better understanding
Loading…
comment
Content-Based Filtering for Internship Recommendations Without User Ratings—Is It Feasible?
Be aware of GenAI tools generated contents regardless of reputation of answer provider. See this screenshot I could generated super similar answer using ChatGPT 4o
comment
Can LSTM be used for non time series data?
Can you explain if there is (significant) correlation between features, thus data records can be treated as sequential and like time-series data ? correlation here means mathematically this term $\beta_2x_{2}$ with feature $x_{2}$ and coefficient $\beta_2$ I would like to learn how from initial assumption "If the packets are sorted in time" we can conclude being sequential
comment
Can LSTM be used for non time series data?
I did not know that in regression model: $y_i=\beta_1+\beta_2x_{i,2}+\beta_3x_{i,3}+\beta_4x_{i,2}x_{i,3}+\epsilon_i,$ in case that if $\beta$ is significant and therefore data is sequential. Any reference? Did you see this post?
revised
How can I ensure about my R^2 score?
Edit and grammar correction in question body for better understanding
Loading…
revised
predictive modelling using Random Forest
Keep the codes\arguments inside of the backticks for better understanding
Loading…
comment
What is the best practice (state-of-the-art) to identify ML-based model learning if it is over/under-fitted or good fit (not diagnosing bad fit)?
I hope this post is not duplication and the most closet post was Overfitting - how to detect it and reduce it? for R language but also the answer was limited focusing on how to choose a good validation set and check the results which is again treatment
revised
Loading…
comment
comment
What is the best practice (state-of-the-art) to identify ML-based model learning if it is over/under-fitted or good fit (not diagnosing bad fit)?
I just find close posts with title of: "Overfitting detection" for this topic but they did not cover best practice to identify but they check cross validation some treatment solutions using less flexible and finally consider Can we use a model that overfits?.
suggested
Approve revised
What is the best practice (state-of-the-art) to identify ML-based model learning if it is over/under-fitted or good fit (not diagnosing bad fit)?
edited title; added 51 characters in body
Loading…
comment
What happens if one uses the forest-based predictive models with a single tree or estimator for 1D time data?
Theoretically we know, "... Gradient boosting (along with any tree-based method) can be used to find relative feature importunes (based on how much error is reduced after each split)", which is not case for 1D time data as my sample sub-datasets are. Thus I could use
n_estimators=1 technically and theoretically ! in the worst scenario there is no further tree's result to get average of it. Right? but question is it still make sense to use such a model? revised
Loading…