Self-Study Data Science

As a data science consultant, lots of people interested in getting into data science have contacted me for guidance on how to get into the field of data science. This article will discuss the recommended topics that one has to study to build essential skills in data science.

The topics presented here, if studied thoroughly, will provide the minimum background needed to start doing data science. This curriculum could also be used for designing an introductory college-level course in data science.

Keep in mind that knowledge acquired from courses alone will not make you a data scientist. Course work has to be accompanied by a capstone project or an internship. Kaggle competitions can be used for capstones, as they provide an opportunity to work on real-world data science projects.

1. Math Basics

(I) 

Most machine learning models are built with a data set having several features or predictors. Hence familiarity with multivariable calculus is extremely important for building a machine learning model. Here are the topics you need to be familiar with:

• Functions of several variables
• Derivatives and gradients
• Step function, Sigmoid function, Logit function, ReLU (Rectified Linear Unit) function
• Cost function
• Plotting of functions
• Minimum and Maximum values of a function

(II) Linear Algebra

Linear algebra is the most important math skill in machine learning. A dataset is represented as a matrix. Linear algebra is used in data preprocessing, data transformation, and model evaluation. Here are the topics you need to be familiar with:

• Vectors
• Matrices
• Transpose of a matrix
• The inverse of a matrix
• The determinant of a matrix
• Dot product
• Eigenvalues
• Eigenvectors

(III) Optimization Methods

Most machine learning algorithms perform predictive modelling by minimising an objective function, thereby learning the weights that must be applied to the testing data in order to obtain the predicted labels. Here are the topics you need to be familiar with:

• Cost function/Objective function
• Likelihood function
• Error function
• Gradient Descent Algorithm and its variants (e.g., Stochastic Gradient Descent Algorithm)

2. Programming Basics

Python and R are considered the top programming languages for data science. You may decide to focus on just one language. Python is widely adopted by industries and academic training programs. As a beginner, it is recommended that you focus on one language only.

Here are some Python and R basics topics to master:

• Basic R syntax
• Foundation R programming concepts such as data types, vectors arithmetic, indexing, and data frames
• How to perform operations in R including sorting, data wrangling using dplyr, and data visualisation with ggplot2
• R studio
• Object-oriented programming aspects of Python
• Jupyter notebooks
• Be able to work with Python libraries such as NumPy, pylab, seaborn, matplotlib, pandas, scikit-learn, TensorFlow, PyTorch

3. Data Basics

Learn how to manipulate data in various formats, for example, CSV file, pdf file, text file, etc. Learn how to clean data, impute data, scale data, import and export data, and scrap data from the internet. Some packages of interest are pandas, NumPy, pdf tools, stringr, etc. Additionally, R and Python contain several inbuilt data sets that can be used for practice. Learn data transformation and dimensionality reduction techniques such as covariance matrix plot, principal component analysis (PCA), and linear discriminant analysis (LDA).

4. Probability and Statistics Basics

Statistics and Probability is used for visualisation of features, data pre-processing, feature transformation, data imputation, dimensionality reduction, feature engineering, model evaluation, etc. Here are the topics you need to be familiar with:

• Mean
• Median
• Mode
• Standard deviation/variance
• Correlation coefficient and the covariance matrix
• Probability distributions (Binomial, Poisson, Normal)
• p-value
• Baye’s Theorem (Precision, Recall, Positive Predictive Value, Negative Predictive Value, Confusion Matrix, ROC Curve)
• A/B Testing
• Monte Carlo Simulation

5. Data Visualisation Basics

Learn essential components of a good data visualisation. A good data visualisation is made up of several components that have to be pieced up together to produce an end product:

a) Data Component: An important first step in deciding how to visualise data is to know what type of data it is, e.g., categorical data, discrete data, continuous data, time-series data, etc.

b) Geometric Component: Here is where you decide what kind of visualisation is suitable for your data, e.g., scatter plot, line graphs, bar plots, histograms, Q-Q plots, smooth densities, boxplots, pair plots, heatmaps, etc.

c) Mapping Component: Here, you need to decide what variable to use as your x-variable and what to use as your y-variable. This is important, especially when your data set is multi-dimensional with several features.

d) Scale Component: Here, you decide what kind of scales to use, e.g., linear scale, log scale, etc.

e) Labels Component: This includes things like axes labels, titles, legends, font size to use, etc.

f) Ethical Component: Here, you want to make sure your visualisation tells the true story. You need to be aware of your actions when cleaning, summarising, manipulating, and producing a data visualisation and ensure you aren’t using your visualisation to mislead or manipulate your audience.

Important data visualisation tools include Python’s matplotlib and seaborn packages, and R’s ggplot2 package.

6. Linear Regression Basics

Learn the fundamentals of simple and multiple linear regression analysis. Linear regression is used for supervised learning with continuous outcomes. Some tools for performing linear regression are given below:

Python: NumPy, pylab, sci-kit-learn

R: caret package

7. Machine Learning Basics

a) Supervised Learning (Continuous Variable Prediction)

• Basic regression
• Multi regression analysis
• Regularised regression

b) Supervised Learning (Discrete Variable Prediction)

• Logistic Regression Classifier
• Support Vector Machine (SVM) Classifier
• K-nearest neighbor (KNN) Classifier
• Decision Tree Classifier
• Random Forest Classifier
• Naive Bayes

c) Unsupervised Learning

• Kmeans clustering algorithm

Python tools for machine learning: Scikit-learn, Pytorch, TensorFlow.

8. Time Series Analysis Basics

Use for a predictive model in cases where the outcome is time-dependent, e.g., predicting stock prices. There are 3 basic methods for analysing time-series data:

• Exponential Smoothing
• ARIMA (Auto-Regressive Integrated Moving Average), which is a generalisation of exponential smoothing
• GARCH (Generalized Auto Regressive Conditional Heteroskedasticity), which is an ARIMA-like model for analyzing variance.

These 3 techniques can be implemented in Python and R.

9. Productivity Tools Basics

Knowledge on how to use basic productivity tools such as R studio, Jupyter notebook, and GitHub, is essential. For Python, Anaconda Python is the best productivity tool to install. Advanced productivity tools such as AWS and Azure are also important tools to learn.

10. Data Science Project Planning Basics

Learn basics on how to plan a project. Before building any machine learning model, it is important to sit down carefully and plan what you want your model to accomplish. Before delving into writing code, it is important that you understand the problem to be solved, the nature of the data set, the type of model to build, how the model will be trained, tested, and evaluated. Project planning and project organisation are essential for increasing productivity when working on a data science project.

I hope you found this useful.

I write about Machine Learning and Data science. If any of those topics interest you, read more here and follow me on LinkedIn && Twitter. ??