Questions tagged [overfitting]
Use this tag for questions related to overfitting, which is a modeling error (especially sampling error) where instead of improving model fit statistics, replicable and informative relationships among variables reduce parsimony, and worsens explanatory and predictive validity.
381 questions
2
votes
1
answer
77
views
How to determine the optimal number of training epochs when validation loss stabilizes but does not increase?
I’m training a CNN (DenseNet169) for a medical imaging task with ~12,000 training samples using fine-tuning (pretrained on ImageNet).
I monitor both training and validation loss/accuracy. What I see ...
- 331
5
votes
1
answer
107
views
How to measure that my dataset is good for the training?
I wanted to train a model for this dataset. the Inputs dataset is here:https://drive.google.com/file/d/1bbMa7auwYjYxyCB72UMBNv5kaojqV7WH/view?usp=sharing and the outputs dataset is here:https://drive....
- 345
4
votes
1
answer
91
views
My neural network doesn't overfit my dataset and my loss won't go down from a certain range
So I made a neural network from scratch, and it seems that my loss doesn't change at all; that's my train code.
...
- 345
8
votes
1
answer
312
views
What does it mean when validation loss increases over several epochs?
I'm training an LSTM model to predict a stock price. This is what I do with my model training:
...
- 219
8
votes
2
answers
384
views
What is the conclusion from this Accuracy / Loss plot for Train and Validation
What is the conclusion from this Accuracy / Loss plot for Train and Validation ?
It seems, that the best results for Validation are after few (5) epochs.
Also I'm not comfortable how the Loss and ...
- 209
1
vote
0
answers
56
views
Overfitting Problem - Medical Image Classification with Transfer Learning (Keras)
Short version: 100% training accuracy, 75-79% testing accuracy
Long version:
I'm a data science noob and my project is to create an ensemble model of 3 to classify retinal fundus (eye) images to 6 ...
- 11
4
votes
3
answers
139
views
Can cross validation for tuning and LOO for evaluation on the exact same dataset cause bias?
I read two articles by the same guy where he uses the whole dataset for hyperparameter optimisation using with CV and then evaluates the model with the best hyperparameters using leave one out on the ...
2
votes
0
answers
83
views
Overfitting problem of convolutional neural network; how to fix?
So I've been working on this convolutional neural network but my accuracy is stuck at 62% without improving and I'm afraid I'm in rather severe situation with the overfitting issue. I've been trying ...
- 21
6
votes
1
answer
180
views
400 instances dataset XGboost, is my model overfitting?
Im working on a regression problem with 400 samples and 7 features, to predict job durations of machineries from historical data. Im using XGboost and (90,10) split works better than (80,20) split. Is ...
- 61
2
votes
0
answers
64
views
Low Dice Score (0.40) in satellite image Segmentation using UNET(Buildings)
I am working on a building roof segmentation task using satellite images, but I am struggling to improve my Dice loss from 0.40. I have tried multiple approaches including:
Different U-Net variants (...
- 21
0
votes
0
answers
52
views
Wouldnt best_param by Random Search CV most likely overfit
I am using RandomizedSearchCV to tune hyperparameters for a RandomForestClassifier and I am concerned about overfitting using rf_random.best_estimator_. (Estimator ...
- 35
1
vote
0
answers
103
views
How do I correctly train with new incoming data sets?
I want to train a deep neural network in a supervised fashion solving a regression task. For the training I can generate data on my own using a certain distribution. Unfortunately the data is very ...
1
vote
3
answers
93
views
Is this model overfitting? My dataset is price for cars
Is this model are overfitting, for what i know is if the difference between training and val loss are high, the model is overfitting and i think this difference is not that high but i dont know. And i ...
1
vote
0
answers
98
views
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'm exploring several ML models for in-sample forecasting task. I'm wondering if there is a straightforward way to identify/detect the good/bad learning.
Classic approach is as it is used for deep ...
- 645
0
votes
0
answers
58
views
What happens if one uses the forest-based predictive models with a single tree or estimator for 1D time data?
I'm experimenting conformal prediction over high-frequent time data using following forest-based regression models for an in-sample forecasting task The size of uni-variate (1D) time-series data is <...
- 645