What could go wrong?

LOL. 😂

Plus the 3 #datascience books that helped me learn #stats the most. 🧵

#rstats Image
I’m not saying you need to be an expert in advanced calculus to do machine learning…

BUT, there is a big difference between someone that does vs someone that does NOT have a good foundation in stats when it comes to getting & explaining business results.
My thought process back in the day was to obtain a great foundation in stats and machine learning at the same time.

So here’s what helped me. I read a ton of books.

Here are the 3 books that helped me learn data science the most...
1. R for Data Science (Wickham & Grolemund) r4ds.had.co.nz
2. Introduction to Statistical Learning (James, Witten, Hastie, & Tibshirani) statlearning.com
3. Applied Predictive Modeling (Kuhn & Johnson) appliedpredictivemodeling.com
Keep in mind that I’ve read 300+ books on stats, ML, time series, …

But these were the 3 best. Ones I got a ton of applied value out of.
Now you’re probably thinking reading these 3 books will take a long time, and still might not get you the whole way to data scientist.

That’s why I want to help you speed up the process.

So it doesn’t take you 5 years to learn data science (like it did me).
I compiled the top 10 most important skills that helped me learn and get results from data science.

And I put these top 10 data science skills into a FREE 40-minute webinar.

Enjoy!

learn.business-science.io/free-rtrack-ma… Image

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with 🔥 Matt Dancho (Business Science) 🔥

🔥 Matt Dancho (Business Science) 🔥 Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @mdancho84

Nov 30, 2023
90% of data scientists overlook how to design A/B Testing experiments.

4 tips for better experiments: 🧵

#DataScience #ABTesting Image
Tip 1: Include a pre-test

Pretest data is unaffected data before the actual A/B test or Time-based Experiment.

Pre-test is a secret used by Booking(dot)com in their CUPED A/B Test method for reducing variance (and improving decision-making from A/B Test results).
Tip 2: Factor in time to effect

For online conversions, sales effects can take time. Your experiment should factor this impact.

A different technique, called Causal Impact can be more important especially if the conversion is a longer sale-cycle / process.
Read 6 tweets
Nov 27, 2023
Both Bayesian and Frequentist approaches to A/B testing have strengths (and weaknesses).

Here's a quick selection guide with 4 Pros/Cons. 🧵

#Bayesian #Frequentist #MachineLearning #ABTesting Image
💡 4 Reasons for the #Frequentist Approach for A/B testing

1. Fixed Sample Size: Requires pre-determination of sample size. Ideal when sample size cannot change once the test begins.
2. P-values and Confidence Intervals: Provides p-values to infer statistical significance and confidence intervals for parameter estimates.

3. Simplicity: Generally easier to explain and understand for those without a strong statistical background.
Read 8 tweets
Nov 20, 2023
Stop using frequentist approaches for A/B Testing.

Use Bayesian instead.

Bayesian has 5 key advantages: 🧵

#DataScience #Bayesian #Rstats #Python Image
1. Intuitive Interpretation:

Bayesian methods provide results in terms of probabilities.

Bayesian probabilities are more intuitive to understand AND more accurate compared to t-test or linear regression p-values.
2. Incorporation of Prior Knowledge:

Bayesian analysis allows the incorporation of prior beliefs or existing data into the analysis.

This is particularly useful when historical data is available.
Read 8 tweets
Nov 13, 2023
12 mistakes that Data Scientists (and even statisticians) make:🧵

#DataScience #Statistics #DataAnalysis #CommonMistakes #CriticalThinking #DataIntegrity Image
Even seasoned data professionals can fall into traps.

Here are 12 common mistakes and misconceptions in statistics and data science:

1. Correlation vs Causation 🔄

Mistaking correlation for causation is a classic error! Remember, correlation does not imply causation. 🚫
2. P-hacking 🔍
Manipulating or over-analyzing data to find significance can lead to false or misleading results. 🚨

3. Base Rate Fallacy 🧮
Overlooking overall event frequencies and focusing on specific information can lead to incorrect probability assessments. ❌
Read 9 tweets
Nov 12, 2023
Can ChatGPT be used for Time Series?

A thread with #R code.

#rstats
I've been using ChatGPT a lot more. But one question I had is whether or not it could be used for Time Series.

In this thread, we'll:

1. Show that chatgpt can write time series code in R
2. Provide code examples
3. Show the app that ChatGPT built for me 👇 Image
1: Can chatgpt write time series code

At first it made some hallucinations, but I was able to get this working code in about 10 minutes:

First, load in these libraries & data: Image
Read 9 tweets
Oct 21, 2023
90% of data scientists struggle with Price Elasticity and Optimization.

Why?

Outliers.

This is how to save your company. (And your career) 🧵

#datascience #stats #BusinessAnalytics Image
Demand is not static. It's constantly changing. And this costs Data Scientists companies and their careers.

But there's a new technique that can help.

Quantile GAMs.
Quantile GAMs allow us to model the heteroskedasticity in our data.

In other words, when variance is non-constant, our model adapts.

Which in turn increases accuracy and profitability.

And I have good news.
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(