Начиная с 30 марта каждую среду Николай Павлов, Data Scientist @ Azzurro.io, проведёт в Fabrika.space серию воркшопов Data Analysis with R.
Data Analysis with R. включает в себя 8 занятий длительностью по 2 часа, посвященных Data Science.
Расписание:
Каждую среду с 30 марта по 18 мая с 19:00 до 21:00.
План занятий
Introduction to data
R programming language
Observations and variables
Relationship between variables
Population and sample
Dependent and independent variables
Experimental design and sampling methods
Data exploration, visualization and cleaning
Data import, cleaning and manipulations
Scatter plot
Histogram, mean, variance and standard deviation
Box plots, quartiles, median and outliers
Data transformations
Categorical data, contingency tables and bar plot
Probability
Outcome, random process and Law of Large numbers
Disjoint/joint outcomes, addition rule
Independence
Conditional, marginal and joint probabilities
Multiplication Rule
Bayes theorem
Random variables, Expected Value, Variance
Probability distributions: PDF, CDF
Normal distribution
Geometric distribution
Binomial distribution
Statistical Inference
Point estimates
Confidence interval
Hypothesis testing
Type I, type II errors, power
Paired data, different of two means
T-distribution
Inference for categorical data
Regression analysis
Linear regression and least squares (LS)
Conditions for fitting regression line
Residuals analysis, R^2
Interpretation and inference
Multiple regression
Model selection
Logistic regression
Predictive Analytics
Machine learning and Supervised learning
Regression / Classification
Error functions
Linear model
Gradient descent, SGD, mini-batches
Decision Trees, Random Forest, Neural Networks, SVM
Bias-Variance tradeoff, regularization L1/L2
Cross-validation
Hyperparameters tuning
BigData, R and Apache Spark
Resilient Distributed Datasets (RDD)
Map-Reduce
SparkR, Data Frame operations
Machine Learning in Spark
Необходимо иметь при себе ноутбук с предустановленным языком R и IDE R-Studio.
Регистрация
+38 (068) 100 11 44
+38 (099) 502 32 46
[email protected]
1 коментар
Додати коментар Підписатись на коментаріВідписатись від коментарів