Shadowing Practice: A/B Testing in Data Science Interviews by a Google Data Scientist | DataInterview - Learn English Speaking with Video

If you are preparing for a data science interview, A-B testing is a must-know concept.

Whether it's for Google, Meta, and Uber, and so forth, A-B testing is a very popular topic in interview questions

because data scientists in those companies will use A-B tests as a way to figure out whether the change

that they have made on those platforms are due to random chance or because of the actual change that they have implemented. And

so what we're going to do in this video is I'll

provide a walkthrough of an A-B test based on real life example

and along the way I'll pepper in some hints that can be helpful in terms of acing interview questions on A-B tests.

Hey everyone, I'm Dan, the founder of DataAntiv.com, ex-Google and PayPal data scientists.

In this video, we're going to do a deep dive on A-B testing based on a real life example

and what I will do is I'll walk through the procedure of setting up an A-B test

and I'll talk about a couple of things that you should definitely mention whenever you're walking through an A-B testing case.

Now let's get started.

The first thing that you need to realize is that when you are walking through the A-B testing procedure, there are essentially seven steps that you need to consider.

The first thing to do is basically understanding the problem, the problem statement.

This is where you try to make sense of the case problem

that you need to solve by asking clarifying questions to the interviewer

and also figuring out what is the success metric and what is the user journey.

And we will definitely do a deep dive on this topic in a second.

The second thing is that you want to define your hypothesis testing.

And what this basically means is that you set up what your null hypothesis and alternate hypothesis is, and you're going to set up some parameter values for your experiment,

such as the significance level and statistical power.

The third step is designing the experiment itself.

And so this is where you talk about what is the randomization unit

and which user type you're going to actually target for this experiment

and various other things that you definitely need to consider when you're designing the experiment.

The next step is to run the experiment itself.

And this is where you need to think about the instrumentation

that is required to actually collect the data and analyze the result.

Now, once you've collected the data, the next thing that you need to do, even before you actually interpret the result and decide to launch,

is basically do some sanity check or validity checks because

if your experiment design was flawed or if there was some bias that was implemented into the data collection itself,

then you have flawed result and so you might end up making poor decisions.

So this is where doing sanity check even before you think about the interpretation and launch decision is very crucial.

Once you have done the sanity check, the next step is to basically interpret the result in terms of what is the lift that you saw,

the p-value encompassed interval, And lastly, now that you have the statistical result along with the business context,

this is where you make a decision in terms of whether you're going to launch the change or not.

Now with all of the steps covered, now let's do a deep dive on an actual case problem.

Suppose

that an online clothing store called Fashion Web Store wants to

test a new ranking algorithm to provide products more relevant to customers.

How would you design an experiment?

So in order to tackle this problem that requires A-B testing, we want to first of all understand the nature of this product.

So we know that this is an e-commerce store that sells goods such as basically clothing goods such as clothes, shoes, bags, and other type of merchandise.

And what this store uses are basically a product recommendation system, an algorithm, where once the user searches some keyword,

let's just say clothing for instance, it's going to generate a result with products that could be relevant for the customer based on, let's just say, their profile,

transaction history, and so forth.

And so what we want to test in this example is that if we change the recommendation system, maybe we're providing more relevant products to the users.

Thereby, it should boost the revenue sales of this e-commerce store.

So we have some general sense about what this problem involves.

And one thing I want to mention is that I've worked with a number of clients on A-B testing interview questions.

One thing I've noticed is that clients will often skip this problem statement part and jump right into basically the statistical methodology,

proposing what the experiment design is.

You don't want to engage the interview question that way.

You always want to start with the business context and then segue over to the statistical methodology.

Now, as you're fleshing out the business goal of this problem one thing

that you want to do is basically clarify what is the user journey

or the product experience that you're trying to change

so this e-commerce store has a following user funnel a user will visit meaning they land on this landing site

and then they will search an item

which basically produces a result based on the recommendation

and then their a user is going to browse a couple items

and then eventually they might click an item and then they'll purchase it

which is the ultimate success that we seek to achieve through the change.

Now thinking about the user journey is important because later down the road

when you're thinking about what is the success metric

and what is the target user population you want to think

about at what stage do you want to consider the user to be participant for the experiment

and this is something we're going to talk about very soon.

Now, one pro tip I definitely want to mention is that when you have an A-B testing interview round, whether it's for meta, Google, go through the individual product,

basically the core product, the core features of that platform, and then create an outline of what that user journey is.

Once you have done this, this is going to be really helpful

when you are in the actual interview setting and you're asked to design an A-B test based on a particular product.

Once you have establish the user journey the next thing to do is basically define the success metric

so what is it that you need to move in order to basically be certain

that the change that you're applying is actually better for the platform overall now

when you think about the success metric you want to think about a couple qualities

so the first thing that you need to think about is

that is it measurable

so is it the type of user behavior you can actually collect through the instrumentation

or the platform and the next quality you want to think about is is your metric attributable

and what this basically means is that can you establish a clear linkage

that the cause the treatment

that you applied to the platform is has led to the effect

that you saw the change in the metric in this case the next quality

that you want to think about is is your metric sensitive

and you essentially have metric that serves as a proxy

that yes the there is a genuine difference in terms of

the user experience in terms of the old algorithm versus a new algorithm

and so statistically speaking you want to find a metric

that has low variability

and a bad metric for in this case let's just say time spent on the website

so the total time spent on the on the website for a given user might have high variability and

so you cannot clearly tell whether there is an actual difference

in terms of how the user is engaging the e-commerce given

that there are some underlying changes to the ranking algorithm the fourth thing is

that a b experiments needs to be very quick it's very

it's a very iterative process as a way to improve a product very quickly

and so you want to ensure that what you're measuring is timely.

You don't want to wait weeks and months to observe a user behavior

100

and then eventually make the change because that's a very costly way of running an experiment.

101

So you want to think about what is that short-term behavior that can serve as a proxy for long-term desired behavior.

102

So based on these four qualities,

103

ultimately the success metric that we want to use for this case is the revenue per day per user.

104

Now once you've established the problem statement clearly, the next thing to do is establish your hypothesis testing.

105

So this is where you state your null hypothesis and alternative hypothesis.

106

So the null hypothesis in this case is going to be

107

the average revenue per day per user between the baseline and the variant ranking algorithms are the same.

108

And the alternative hypothesis in this case is going to be that the

109

average revenue per day per user between the baseline and the variant ranking algorithms are different.

110

Once you've state the hypothesis statements, the next thing that you want to do is you want to set the significance level, or your alpha in this case.

111

So the significance level is basically the decision threshold.

112

If the probability of observing a particular event is very low, then it is deemed statistically significant.

113

And so in this example case, the significance level that we want to set is 0.05.

114

And that's the usual value that is often set in online experiments the next value

115

that we want to set is this physical power the statistical power usually is set as 0.80

116

and what 0.80 basically means is is the 80 probability of detecting effect given

117

that the alternative hypothesis is true

118

and lastly what you want to set is your practical significance basically the minimum detectable effect

119

and typically for a large online platform with millions of users

120

the mde is one percent lived once you've set up the hypothesis testing the next thing

121

that you want to do is design the experiment itself

122

so the first thing

123

that you want to consider as you design the experiment is

124

you want to consider what is the randomization unit in this

125

case we're going to randomly assign at the user level basically randomly assign them in terms of the control group

126

or the treatment group once you have defined the randomization unit you you need to think about

127

which population of the users you want to target.

128

And this is something that I talked about earlier in terms of the user funnel, right?

129

So you have a user that visits, you have a user that searches, they browse, they will visit an item and then they'll eventually purchase

130

so at what level do you want to allow the user to participate in the experiment

131

so in this case we actually want to target users who have actually started searching something why

132

because this is where the algorithm actually kicks in

133

and then they're actually exposed to the treatment in this case

134

which is either the the old algorithm or the new algorithm the third thing

135

that you want to define is your sample size.

136

The general rule of thumb is the following formula, which is that n is approximately equal to 16 times the variance divided by the delta square,

137

where delta essentially represents the difference of the key metric between the treatment and the control.

138

And this formula is based on the assumption that your significance level is 0.05 and your statistical power is 0.80.

139

Once you have determined your sample size, the next thing that you want to do is determine the duration of your experiment.

140

And the typical duration of the experiment is going to be one to two weeks.

141

You don't want to run the experiment less than one week

142

because you want to account for the day of the week effect, meaning

143

that there could be some underlying difference in terms of how the user engages the website during the weekdays versus the weekends.

144

Once you have designed the experiment, the next thing that you want to do is you want to run the experiment

145

and this is where you use instrumentation some experiment platforms as a way to collect the data

146

and track your result now it is very important that

147

while you're running the experiment you do not peak at the

148

p-value meaning you don't make any decision in terms of whether you're going to launch

149

or not while the experiment hasn't been completed yet the reason is

150

that when you're peaking there's a chance that

151

when you have low sample size for instance there's a lot of variability in terms of where

152

that lift goes and so you might falsely conclude that there is an underlying difference when there isn't.

153

So once you have determined what the experiment time is, given your statistical power and your sample size, you have to ensure that you wait it out.

154

Otherwise, you're going to increase the chance that you falsely reject the null hypothesis given that it is actually true.

155

After you have run the experiment, the next thing that you want to do is you want to perform validity checks.

156

So this is where you conduct sanity checks including

157

instrumentation effect or are there any bugs or glitches

158

that could potentially affect the experiment result another potential issues

159

that you want to look at are external factors

160

so maybe you run the experiment during the holiday or

161

when competition launched something very important

162

or it could be some general economic conditions like covid

163

or recessions and when you run the experiment when some external disruptions happen this could potentially impact your experiment result.

164

So ideally you want to run an experiment that avoids periods like this.

165

The next thing that you want to do is you want to check for selection bias.

166

And what this basically means is that you want to assume that the underlying distribution between the control and the treatment group,

167

even before they're exposed to the treatment condition in this case, is that they're homogeneous.

168

And so one way to confirm that the distributions are the same is to basically run an AA test.

169

The next thing that you want to check is a sample ratio mismatch.

170

What it basically means is that if you're randomly assigning a user into the control or the treatment group, then out of all the participants of the experiment,

171

50% of them should be in the control and 50% of them should be in the treatment.

172

But there are cases where because of some flaws with the randomization algorithm, that the ratio is actually not 50 to 50%.

173

It might be 49 to 51%.

174

So in order to ensure whether this could pose a potential issue later down the road,

175

you want to use a chi-square test as a way to ensure that the ratio between the two samples is sound.

176

The next item you want to check is a novelty effect.

177

What this basically means is that if you made some change to the website itself,

178

a user might have reacted simply because there's a novelty behind being exposed to something new.

179

One way to detect a novelty effect is basically look at it by user segment.

180

Look at the underlying difference of the success metric in terms of new visitors versus the recurrent visitors.

181

And if you see that there is a change between the two, then there is a presence of novelty effect.

182

So you may actually want to run the experiment where you

183

segment it by the new visitor group versus the recurrent group first.

184

You've conducted the validating check and there's no issue with the experiment experiment, now you can actually interpret the result.

185

And when you interpret the result, you want to look at the direction of the success metric, is a lift negative or positive?

186

And you want to consider the p-value because

187

that helps you establish whether the lift that you saw is statistically significant or not.

188

And you also want to consider the confidence interval in this case.

189

So based on the experiment that we ran for this example, what we see is

190

that the average revenue per day per user in the control group is $25 whereas in the treatment group is $26.10.

191

This produced a falling lift.

192

So in terms of the absolute difference is $1.10.

193

In terms of relative lift, it is an increase of 4.4%.

194

And the lift we saw is statistically significant because we see

195

that the p-value 0.001 is less than the statistical significance at 0.05.

196

And we also see that the confidence interval at the significance level of 0.05 is between 3.4 and 5.4% lift.

197

So the initial interpretation based on what we see is the following.

198

There is statistical significance to reject the null hypothesis and conclude

199

that the average revenue per day per user between the baseline and the variant ranking algorithms are different.

200

With this result, we can now consider whether we want to launch or not.

201

Now when you think about whether you want to launch it or not, and in this step there are three factors that you want to consider.

202

The first factor is the metric trade-off.

203

So you might have a case where the success metric might have improved, but the guardrail metrics or the secondary metrics might have declined.

204

And so you have to think about what are the pros

205

and cons of launching it considering that the guardrail metrics might have declined.

206

The next factor you want to consider is the cost of launching.

207

If you see that the cost of building this out, basically rolling out to all of the users, and the cost of maintaining this change is highly costly,

208

then maybe this isn't something that you want to actually launch.

209

The last factor you want to consider is the risk of committing false positives, or your type 1 error rate.

210

For instance, if you falsely conclude that there is an effect when there isn't, and you've made a change then it might have a negative

211

consequence to the user you might end up providing poor experience to users

212

and they might turn and ultimately

213

that is going to negatively affect the bottom line of the product

214

so these are three important business factors

215

that you want to consider now you want to kind of go back to the interpretation of the result

216

that you got from the experiment and along with the business context

217

and the statistical result ultimately you want to decide whether you're going to launch it or not

218

so what we're going to do is we're going to look

219

at a couple example cases where we'll look at a possible range of lift along with the confidence intervals

220

and then think about what is a sound decision based on the result that we have seen.

221

So in this first case what we see is

222

that the lift is placed at a positive value

223

but it's still less than practical significance and you also see

224

that the lower bound

225

and the upper bound of the confidence interval are less than the practical significance in this case

226

which is going to be

227

positive one percent so in this case what you want to consider is

228

that maybe you want to change the algorithm

229

or scrap the idea all of it together in the next case what we see is

230

that the lift and the bounds the confidence interval is uh practically significant

231

so this provides a strong support

232

that we should make a launch in the third case we see

233

that the entire interval is less than zero

234

so it's in the negative territory in this case you want to consider perhaps iterating the idea or scrapping it all together.

235

And the next example is you have a positive direction in the expected lift, but then the bounds are in the negative territory and the positive directory.

236

And you can see that it's a very wide bound.

237

Now, what is very important to note is that the upper bound as interval is practically significant.

238

It's greater than 1%.

239

So there is still some likelihood that you might see a lift with practical significance.

240

So what you want to consider in this case is actually to rerun the experiment with increased statistical power

241

and this is going to help improve the precision of that lift that you're seeing.

242

And in the last case you see that the lift itself you see

243

that the expected lift is practically significant but you see

244

that the lower bound of this is not practically significant but it's still in the positive territory.

245

And the best thing to do in this case is to

246

rerun the experiment with increased statistical power just to be absolutely sure that the underlying change is practically significant.

247

So based on all of these considerations in terms of the business context, along with various statistical outcomes.

248

Ultimately, the decision you want to make is

249

that you want to launch this new algorithm as a way to provide a more relevant product recommendations to the users, thereby improving revenue overall.

250

So there you have it, guys.

251

This is the end-to-end process of how to walk through an A-B test, how to actually address an A-B testing interview question.

252

I hope you found this video really helpful

253

if you need any help in terms of mock interview coaching

254

or courses that comes with a b testing courses

255

and various business case problems and slack community access definitely check out data ntv.com

256

and if you have any questions along the way feel free to drop a comment down below

257

or feel free to send me an email at dan at data ntv.com i'll see you in the next video

✨ Recomendar Video

245 oraciones

¿Qué es la Técnica de Shadowing?

Shadowing es una técnica de aprendizaje de idiomas respaldada por la ciencia, desarrollada originalmente para la formación de intérpretes profesionales y popularizada por el políglota Dr. Alexander Arguelles. El método es simple pero poderoso: escuchas audio en inglés nativo y lo repites en voz alta de inmediato, como una sombra que sigue al hablante con solo 1-2 segundos de retraso. A diferencia de la escucha pasiva o los ejercicios de gramática, el shadowing obliga a tu cerebro y músculos de la boca a procesar y reproducir simultáneamente patrones de habla reales. Las investigaciones muestran que mejora significativamente la precisión de la pronunciación, la entonación, el ritmo, el habla conectada, la comprensión auditiva y la fluidez al hablar, convirtiéndola en una de las metodologías más efectivas para la preparación del IELTS Speaking y la comunicación en inglés en el mundo real.

Técnica de shadowing: lee la guía completa paso a paso →

Shadowing Practice: A/B Testing in Data Science Interviews by a Google Data Scientist | DataInterview - Learn English Speaking with Video

QA Automation Interview Questions and Answers

Job Interview English 🇬🇧 | Real Conversation Practice (B1–B2 Level)

Learn how to SOLVE a data analytics case study problem

FULL Data Science Mock Interview | SQL, Statistics, Business

Netflix product manager mock interview and tips (with ex-Netflix Lead PM)

Digital Marketing Specialist Interview Questions and Answers

I'm an ex-Google interviewer. You're doing LeetCode wrong.

How AI Is Ruining Job Interviews

Phone Interview with a Recruiter

Distributed Task Scheduler: System Design Interview (Stripe & Amazon Offers)

Google Interview Tips That Actually Get You Hired

Top 5 Insider Interview Questions Data Analysts Must Master Before Any Interview!

Context & Background

Top 5 Phrases for Daily Communication

Step-by-Step Shadowing Guide

¿Qué es la Técnica de Shadowing?