Let’s say you’re working as a data scientist for a product company, and are tasked with assessing whether a new feature should be implemented.
In simple words, you will test the success of the feature among a small sample of users. And based on the result of that experiment, you will decide whether to release the feature to all users or drop it entirely.
The above experiment is an example of statistical inference, which is the process of drawing conclusions about a population based on a sample.
When conducting research or performing experiments, we generally cannot obtain data from the entire population.
For example, if you wanted to test the efficacy of a vaccine used to prevent a deadly disease, it isn’t feasible to conduct this test on the world’s population. Instead, you must choose a sample from this population and conclude whether the vaccine works based on a subset of people.
This is where statistical inference comes in, and is a must-learn topic for data scientists.
Statistical inference also has applications in fields like manufacturing, social science, product analytics, and finance.
In this article, I will provide you with 5 free online courses to learn statistical inference. After taking these courses, you will be prepared to perform statistical tests in real-world settings and thrive in your data science role.
1. Intro to Inferential Statistics (Udacity)
Udacity’s Intro to Inferential Statistics course is a beginner-friendly learning resource that covers concepts like confidence intervals, hypothesis testing, sampling distributions, and the different statistical tests (T-Test, Chi-Squared Test, ANOVA).
This test assumes that you have some knowledge of descriptive statistics, i.e. you must be familiar with concepts like variable distribution, central tendency, and variability.
If you don’t know how to calculate a dataset’s median or are unable to analyze the relationships between variables in a dataset, then I recommend taking Udacity’s free Intro to Statistics course before moving on to this one.
2. Inferential Statistics by Duke University (Coursera)
Duke University’s Inferential Statistics course is part of the “Data Analysis with R” specialization on Coursera, which is a broader learning path that covers topics like probability, descriptive statistics, and machine learning modeling.
Here’s what you will learn in the inferential statistics course:
- Confidence Intervals
- Central Limit Theorem
- Hypothesis Testing
- Power Analysis
- ANOVA
- Bootstrapping
- Inference for Categorical Data
- Chi-Square Tests
Each topic comes with reading material, practice exercises, and hands-on implementation in R. You will learn not just the theory behind the above concepts, but will also be taught to apply them to real datasets.
This course can be audited for free. This means that you can access all the course lectures and learning resources for free, although you will need to pay to obtain a course certificate.
3. Inferential Statistical Analysis with Python (University of Michigan)
The Inferential Statistical Analysis with Python course is a part of Coursera’s “Statistics with Python” specialization.
The concepts you will learn in this course are very similar to the ones covered by the learning resources provided by Udacity and Duke University, although all the practical sessions are taught with Python.
You will learn to use Python libraries like Statsmodels, Pandas, and Seaborn to conduct statistical tests and interpret their results. You will also learn to perform statistical analysis on real datasets like the NHANES (National Health and Nutrition Examination Survey), which will prepare you for real-world data science applications.
4. AP Statistics (Khan Academy)
Khan Academy’s AP Statistics course teaches you statistics from the ground up. The first few units are focused on beginners and cover descriptive statistics, measures of central tendency, and probability theory.
You can skip these topics if you’d like to dive straight into inferential statistics if you’d like.
Unit 9 onwards covers areas like sampling distributions, confidence intervals, and statistical inference for quantitative and categorical data.
5. Data Science: Inference and Modeling by Harvard University (edX)
The “Data Science: Inference and Modeling” course is part of Harvard’s Data Science Professional Certificate series.
This course covers the principles of statistical inference, the Central Limit Theorem, confidence intervals, and Bayesian statistics.
You will learn through case studies and examples and will be taught to apply the topics covered using R programming. The program requires some prerequisite knowledge in R, data visualization, and probability.
Learning Statistical Inference – Next Steps
After completing some of the above courses on statistical inference, I recommend applying your knowledge to a variety of datasets to put your skills into practice. You can find datasets on sources like Kaggle and the UCI Machine Learning Repository.
If you aren’t already proficient in programming, I suggest brushing up on your Python or R skills. This will allow you to take advantage of the software’s statistical libraries, turning you into a “unicorn data scientist” with a deep knowledge of statistics, machine learning, and coding.
Finally, you might find it beneficial to join statistics communities on platforms like Reddit and StackOverflow. If you get stuck during the learning process, you can always ask questions and participate in forum discussions, strengthening your knowledge of the subject.
