Important concepts on Statistics for Data Science interviews [Part 2 of 4] | PSN Academy
11. Bias: Assumptions made by a predictive algorithm. The error occurred due to bias is called Bias error or Error due to bias.
Low Bias algorithms: k-NN, SVM, Decision trees
High Bias algorithms: Linear regression, Linear Discriminant Analysis (LDA), Logistic regression
Variance: Amount of change in estimate by the learning algorithm 𝐿 on changed training dataset.
Difference: Too large: Model overfitted | Too small: Model underfitted
Low Variance algorithms: Linear regression, Linear Discriminant Analysis (LDA), Logistic regression
High Variance algorithms: k-NN, SVM, Decision trees
12. Coefficient of determination or 𝑹^𝟐
Degree of variability in the target variable explained by the independent variables.
𝑅^2=Explained variation /Total variation
Explained variation (aka Target variation): Independent variables are responsible for.
Unexplained variation: Independent variables are NOT responsible for.
13. Z-score (aka Standard score): A value whose unit of measurement is Standard deviation from the mean.
14. Maximum Likelihood Estimation: A probabilistic framework for finding the values of parameters for which the model has the best fit i.e. optimized.
15. Bayes Theorem
Joint probability: The probability of occurrence of one event with simultaneous occurrence of another event.
Marginal probability: Probability of an event for one random variable irrespective of any other random variable present in the dataset.
Conditional probability: The probability of one event given another event has already occurred.
𝑃(𝑆|𝑀)=(𝑃(𝑀|𝑆)∗𝑃(𝑆))/𝑃(𝑀)
16. KOLMOGOROV-SMIRNOV GOODNESS OF FIT TEST
KS test will say whether the data follows the assumed distribution or not.
17. Confidence Interval: The inference of estimating the true value of parameter by a
pair of estimate values which are constituted an interval in which true value of
parameter expected to lie with certain confidence. The technique of finding
such interval is known as “Interval Estimation”.
18. ANOVA: A test of hypothesis where hypothesis states that the means of populations are equal.
19. POISSON DISTRIBUTION: Estimates the probability of occurrence count of events in a large population over a unit of time.
20. CORRELATION COEFFICIENT: Correlation: Studies the linear relationship between the two or more variables. If a change in the value of one variable is observed alongwith a change in the value of another variable, then these two variables are said to be correlated or there is correlation between these two variables.
----------------------------------------
Chapters:
00:18 What is Bias in Bias-Variance tradeoff?
01:31 What is Variance in Bias-Variance tradeoff?
05:17 What is Co-efficient of determination or R-squared?
05:56 What is Explained variation?
06:10 What is Unexplained variation?
07:46 What is Z-score or Standard score?
11:35 What is Maximum Likelihood Estimation?
14:27 What is Likelihood function in MLE?
14:37 What is Log-Likelihood function in MLE?
16:00 What is Joint probability?
17:02 What is Marginal probability?
17:52 What is Conditional probability?
19:04 What is Bayes theorem?
20:22 What is Kolmogorov Smirnov test?
24:34 What is Confidence Interval?
29:14 What is ANOVA?
33:43 What is Poisson distribution?
38:02 What is Correlation coefficient?
#datascienceinterview #datascienceinterviewquestions #kstest #correlationcoefficient #poissondistribution #anova #confidenceinterval #bayestheorem #mle #maximumlikelihoodestimation #zscore #rsquared #bias #variance #tradeoffs #psnacademy