Suppose your current random forest classifier is facing an overfitting situation. Please state whether increase or decrease the following parameters can potentially help you reduce overfitting? (i) n_estimators (ii) max_features (iii) max_depth (iv) n_jobs
Added by Francisco P.
Close
Step 1
** Show more…
Show all steps
Your feedback will help us improve your experience
Qudsiya Anis and 85 other Intro Stats / AP Statistics educators are ready to help you.
Ask a new question
Labs
Want to see this concept in action?
Explore this concept interactively to see how it behaves as you change inputs.
Key Concepts
Recommended Videos
A random forest is an ensemble learning method that attempts to decrease the bias of decision trees. An infinite depth binary point is mislabeled decision tree can always achieve 100% training accuracy, provided that no outliers are present. K-Means clustering looks to find low-dimensional representation of the observations that explain a good fraction of the variance. Linear SVMs, widening the margin increases the number of observations that violate the margin of the classifier. A natural spline is a regression spline with the additional constraints that the function is required to be linear at the boundaries.
Tanvi G.
Increasing Training Set Size Experiment: Consider the iris dataset for multiclass classification and perform the following steps: 1. Divide the data into 80% training and 20% testing. 2. From the training set, only take 5% of the data and train the supervised learning models (Logistic Regression, Decision Trees, Random Forest, and Naive Bayes) and test them on the test set created in the previous step. 3. Repeat the training again with 10% of the data and keep on adding 5% until you use the whole training set. 4. In every training, test on the 20% of the test set and report the accuracy and F-score of the model. 5. Plot the sample graph for accuracy and F-score as provided below:
Akash M.
QUESTION 3 You create 4 models to predict when a type of MRI machine found in several of your hospitals in your health system will "break down within 30 days". You collect data from your equipment regarding several scanning conditions during optimal, pre-faulty, and faulty conditions. This info comes from multiple machines at various locations. You also collect data on the types of scans administered, as well as several environmental conditions. In order to avoid overfitting this model and thus having inaccurate maintenance schedules (maintenance is costly and MRI outages are a patient safety concern), you can do which of the following: A. Add 20 additional variables to your models since more variables = more flexibility and less overfitting. B. Reduce the size of your training set in relation to the validation and test sets. Too much training data will cause overfitting. C. Reduce the number of variables in one or more of your models. This may remove noise and thus mitigate overfitting. D. Only select data from the MRI machine that has the most patients. That way you don't overfit your model to a machine that is rarely used. E. Create 10 additional alternative models. The more models you evaluate, the less bias to an overfit model you will likely encounter.
Ameer S.
Recommended Textbooks
Elementary Statistics a Step by Step Approach
The Practice of Statistics for AP
Introductory Statistics
Transcript
Watch the video solution with this free unlock.
EMAIL
PASSWORD