Alexey Grigorev
  • Member for 7 years, 6 months
  • Last seen more than a week ago
6 answers
27 votes
9k views
Deep learning basics
14 votes

Neural Networks and Deep Learning by Michael Nielsen. The book is still in progress, but it looks quite interesting and promising. And it's free! Here's the link. There are only 5 chapters so far, and ...

View answer
4 answers
34 votes
49k views
Do Random Forest overfit?
12 votes

You may want to check cross-validated - a stachexchange website for many things, including machine learning. In particular, this question (with exactly same title) has already been answered multiple ...

View answer
4 answers
18 votes
13k views
Dimensionality and Manifold
Accepted answer
34 votes

What is a dimension? To put it simply, if you have a tabular data set with m rows and n columns, then the dimensionality of your data is n: What is a manifold? The simplest example is our planet ...

View answer
1 answers
117 votes
325k views
How to get correlation between two categorical variable and a categorical variable and continuous variable?
115 votes

Two Categorical Variables Checking if two categorical variables are independent can be done with Chi-Squared test of independence. This is a typical Chi-Square test: if we assume that two variables ...

View answer
1 answers
3 votes
574 views
XGBoost in final model what does "yes=3,no=4" mean?
Accepted answer
2 votes

I assume it should read as follows: if condition is true, go to branch 3 otherwise go to branch 4 I am only guessing, but seems plausible

View answer
5 answers
4 votes
7k views
Best approach for this unsupervised clustering problem with categorical data?
-1 votes

You first need to convert this data into some numerical representation, and then you can use clustering. One of such ways is applying TF-IDF weighting to tags, and then calculate the cosine ...

View answer
2 answers
0 votes
225 views
is there any whatever2vec to generate one vector for one document?
1 votes

Yes, there are such things, but they aren't really related to word2vec. First of all, Bag of Words is the old "something2vec" approach of representing documents in a vector space. Of course, the ...

View answer
3 answers
4 votes
3k views
When does it makes senses to use Dot-Product as similarity measure instead of Cosine?
1 votes

The relation between dot product and cosine is similar to the relation between covariance and correlation: one is normalized and bounded version of another. In my experience usual dot product is ...

View answer
2 answers
4 votes
3k views
Creating Domain specific Question Answering Systems
1 votes

You should check the Allen AI competition on kaggle: https://www.kaggle.com/c/the-allen-ai-science-challenge In short, the typical approach people took there was similar to what you're suggesting: ...

View answer
1 answers
3 votes
106 views
Tokenizing words of length 1, what would happen if I do topic modeling?
Accepted answer
2 votes

The libraries usually exclude 1-length tokens and tokens with no alpha-numeric characters because typically they are noise and do not have any descriptive power. That is, these tokens are usually not ...

View answer
1 answers
3 votes
2k views
Handling categorical features in Factorization Machines algorithm - Feature Hashing vs. One-Hot encoding
Accepted answer
5 votes

I decided to expand a bit on my comment and make a full answer. So the reason why somebody may say that performing the hashing trick can destroy interactions is because it may map different features ...

View answer
2 answers
5 votes
852 views
How are clusters from DBSCAN sometimes non-convex?
1 votes

I think it's non-convex because the particular cluster assignment you get when applying DBSCAN depends on the order you traverse the data. Let's try to illustrate it with an example. Consider this ...

View answer
2 answers
10 votes
1k views
Implementing Complementary Naive Bayes in python?
Accepted answer
5 votes

Naive Bayes should be able to handle imbalanced datasets. Recall that the Bayes formula is $$P(y \mid x) = \cfrac{P(x \mid y) \, P(y)}{P(x)} \propto P(x \mid y) \, P(y)$$ So $P(x \mid y) \, P(y)$ ...

View answer
3 answers
2 votes
2k views
Reading Persian Characters in R
1 votes

make sure the files are saved in UTF try Sys.setlocale("LC_ALL", locale_code) and have a look at the documentation of this function

View answer
3 answers
4 votes
253 views
Could one algorithm fetch keywords from texts of different natural languages?
2 votes

So my question is: Would I need to create separate algorithms for every natural language to be interpreted? Yes, I believe so. But building a model for detecting the used language is not hard: ...

View answer
7 answers
12 votes
1k views
What is an 'old name' of data scientist?
2 votes

Also: "Business Intelligence developer"

View answer
2 answers
9 votes
472 views
How to build a textual search engine?
4 votes

pre-process your documents (some of the steps may be skipped) tokenize remove stop words stem or lemmatize do normalization (e.g. U.S.A. -> USA, météo -> meteo, etc) and orthographic correction ...

View answer
1 answers
8 votes
2k views
Difference between tf-idf and tf with Random Forests
Accepted answer
7 votes

Decision trees (and hence Random Forests) are insensitive to monotone transformations of input features. Since multiplying by the same factor is a monotone transformation, I'd assume that for Random ...

View answer
4 answers
34 votes
15k views
Quick guide into training highly imbalanced data sets
20 votes

Undersampling the majority class is usually the way to go in such situations. If you think that you have too few instances of the positive class, you may perform oversampling, for example, sample 5n ...

View answer
4 answers
4 votes
316 views
Handling huge dataset imbalance (2 class values) and appropriate ML algorithm
2 votes

In addition to undersampling the majority class (i.e. taking only a few NEW), you may consider oversampling the minority class (in essence, duplicating your OLDs, but there are other smarter ways to ...

View answer
2 answers
14 votes
4k views
Analyzing A/B test results which are not normally distributed, using independent t-test
Accepted answer
8 votes

The distribution of your data doesn't need to be normal, it's the Sampling Distribution that has to be nearly normal. If your sample size is big enough, then the sampling distribution of means from ...

View answer
3 answers
4 votes
182 views
Facebook's Huge Database
1 votes

When I worked with social network data, we stoted the "friendship" relation in a database in the table Friends(friend_a, friend_b, ...) with a B-Tree index on (friend_a, friend_b) plus also some ...

View answer
3 answers
26 votes
43k views
Data Science Project Ideas
3 votes

Introduction to Data Science course that is being run on Coursera now includes real-world project assignment where companies post their problems and students are encouraged to solve them. This is done ...

View answer
3 answers
10 votes
814 views
Network analysis classic datasets
4 votes

Maybe you can check here - http://snap.stanford.edu/data/ For each data set you will also see references of the works where they have been used

View answer