Revision

2020

Meta - Research Data Scientist

Math: 1) Questions on linear regression

• Formula
• What to do with outliers (outlier detection, l1 regression)
• Test on parameters
• What problem appears with test if number of parameters tends to infinity (if alpha=5%) 2) Explain Gradient Boosting, SVM

Algo: 1) Remove kth element from end

Open Problem: Design a solution to detect videos with content violation

2021

2022

Meta - Research Data Scientist

Math: 1) Application of Baye’s rule 2) Classification dataset (3 lines) with questions on it 3) Probability question: paquet are sent to a server. Probability of being well sent is 0.6. What is the probability that over 10 paquets, one is not sent well. First not sent and other sent = 0.4 * 0.6^9; Second not sent and other sent: 0.6 * 0.4 * 0.6^8, … => 10 * (0.4 * 0.6^9) = (10, 1)(0.4 * 0.6^9) (10 chooses 1) 4) Statistic: Is a dice faire: Average of N dices is X (2 ?) => Distribution of sum of n dices and then average of n dices (distribution/mean/std) => Which test => What does the p-value represent => Conclusion: is the dice fair? 5) Formula of R^2: 1 - SS_Residual/SS_Total = SS_Explained/SS_Total as SS_Total = SS_Explained + SS_Residual 6) Classification dataset (3 lines) with questions on it (VIF)

Data Manipulation Algo: 1) Two files: one with one word per line and one with on sentence per line. Count how many time each word of the first file appears in the second file. Read in Python + Counter or regex. 2) 3) Given a string of digit, you can use one swap of values to maximise the number.

Open: Design a solution to predict the number of pictures that a new user will post in the next 30 days

Kraken - Data Scientist

…

Akur8 - Data Scientist

Generalized Linear Model:

• Show that MLS and likelihood led to the same result under the normality assumption,
• Questions on logistic regression

CACIB - Algorithmic Trading

• Code: Absolute Maximum Subarray
  Math: Integral of exp(-x^2) between -inf and +inf (use pdf of gaussian distribution)
  Open: Hedge Bond with a portfolio of given Bonds (lasso regression on the performances)\

Meta - Software Engineer Machine Learning

Code 1: Find if a string of only parenthesis is correct (ie parenthesis are correctly balanced)
Code 2: Mouse in a Maze with no vision of the environment. It should find the cheese somewhere in the maze.

• You are given 2 functions :
  • One function Move that take in parameter the direction (up, down, left, right) and return True if the move can be made and False otherwise (if there is a wall for example),
  • One function HasCheese that return True if the current state has the cheese