A (rather funny) Data Science story.

(At Lana's apartment) 

• Lana: I like Roger. I think he could make me happy. (Hypothesis) 
• Steph: By no chance. He’s a professional cheater, I am pretty sure about it. (Another hypothesis) 
• Lana: I am so confused… What's more important: His loyalty to me or his ability to make me happy? (Data Science’s uttermost matter: Are we asking the right question?)
• Steph: Bah… what's happiness anyways? It depends on the eye of the beholder. Instead, cheating is something that is “black and white”: either he cheats on you or not. (Every Data Science question should be specific) 
• Lana: Let’s stick with the basics: What would upset me the most is him flirting with other girls; I will dump him out if I catch him doing that. (Refining the question)
• Steph: I have a great idea! I will ask Gloria about how to determine whether a man is a womanizer or not. (Model selection) 

(WhatsApp conversation)

• Steph: Wussup, Gloria! How do you determine if a man is a womanizer or not? (Consulting to an expert) 
• Gloria: Womanizers do party a lot, and they dress very fancy clothes. Also, they tend to be tall people. Trust me, it's been years of dating a lot of guys, hun. (Domain expertise)
• Steph: That seems pretty reasonable, except for the height; it just sounds crazy! Thanks a lot! (Variable selection) 

(At Lana’s apartment -again-)

• Lana: Steph: have you figured out a way to determine if he’s a womanizer or not? Today, he’s attending his sister's wedding reception, and there's going to be a lot of girls; I don't want to be seen as the “poor victim”. (Business asking for results)
• Steph: Calm down, Lana! I did figure out a way for doing that. You know, Roger is a fancy guy, and does party a lot, so he surely is a womanizer. I knew it, I knew it, I knew it! (Data Insight) 
• Lana: Do you really think that he's a womanizer? I don't think so. I need a firmer ground to stand on. (Skepticism about findings) 
• Steph: I’ve asked Gloria, which is an expert on these matters, and I've got her feedback. Also, I've added some ideas of my own and experiences I’ve heard in the past, so I won't be wrong. (Model training) 
• Lana: have you asked several people about this? Just make sure this is not a thing of one or two people. (Statistical power) 
• Steph: I would bet my honor without any fear on that! In fact, I will go and enter the wedding without authorization, and thus I will demonstrate that my way of thinking is accurate. See ya! (Model validation)

(At Roger’s sister’s wedding)

• Steph: What a filthy animal! Roger is dancing with that girl, and he’s speaking to her next to her ear, feeling so confident and comfy… And look at her! I can't believe it! (Model in production before proper validation) 
• Next chair’s neighbor: She's another sister of him. (Statistical evidence is not enough to reject null hypothesis) 
• Steph: Oops… (Type I error) 

(At Lana’s house -with a XXXL hangover-) 

• Steph: I've got to tell you that life is unfair. (Bad result communication) 
• Lana: Say what??? (Message is unclear for intended audience) 
• Steph: Can't you get my point? I thought Roger was a womanizer, but I've found out he does not really seem to be. (Model not valid for implementing) 
• Lana:  Are you kidding me? I've dumped him once you told me what you thought… (Making wrong, rush-based decisions due to a wrong Data Science process) 

Yikes! (Source: https://letsalldogood.files.wordpress.com/2015/12/yikes-4.png)