@chartosaur: ⚠️ WHEN/WHEN NOT TO USE LOESS Not every dataset wants to be a line chart. Some just want to be left alone as a scatterplot. LOESS is a smoothing method that reveals the trend without forcing jagged lines between every point. Too often, people connect the dots and end up with a zigzag mess. LOESS helps cut through that chaos—highlighting the overall shape of the data, not every bump. ✅ Use LOESS when: • You’ve got a dense, noisy dataset • You’re telling a story, not reporting every fluctuation • The big-picture trend matters more than each data point 🚫 Avoid LOESS when: • Your dataset’s small or thin (it can overfit fast) • Your data has sharp shifts or seasonality • You need clarity and precision—LOESS is trend-first, not point-perfect Also know: It doesn’t love outliers, it’s slower on big data, and it won’t give you an equation. LOESS is for visual storytelling, not exact prediction. Bottom line: Don’t force lines where they don’t belong. Use LOESS when you want clarity over chaos. #Charts #Presentation #Viz #PPT #Excel #Graph #Consulting #Mckinsey #Bain #BCG #Visualization #Slides #Charts #Graphs #Deck #GoogleSlides #LineChart #Education #Data

Blantaya :

Doesn’t this remove a lot of info tho?

nash.rougvie :

What about a rolling average? Feels like this smooths the data too much and doesn’t show how volatile the data actually is

GioGion :

I don't see the purpose, the jagged one is less confusing and provides more immediate information, the second one also "hide" critical discussion point

QuantumCoherence :

Chartosaur :

did LOESS make it better?

You :

Thanks for the reminder Sir Saur

xPercenth :

I mean the dots seam very volitile

Kev Lobs :

but I could see the trend in the first place 😭

Avis🤔🥀 • Friends ∙ Friends :

The first one is better

professor dog :

Break it into quarters and it’ll look much smoother

Yann :

Quick question, what website or app do you use?

Astolfo_Kys :

Or you know you can just stretch the time lane

James :

Nice!

virg09876 :

what the name of the graph and the analysis?

sexy_science :

then put a note that the smoothed line is an approximation

ducminh :

moving average?

tyler :

how to do it ?

andyrufasto :

nice one

marinosierra :

It depends on what the objective is. In general I don't like it, it detracts from important information.

Dr Antler :

I disagree with you this time . First, the first plot looks better to me. 2nd once you put these lines you insert a bias

Oscarconan :

😁

okginger :

😂😂😂

egerex :

bro anyone who is mildly educated in mathematics could see the “trend”. on top of that, there isnt a need for a trend to exist just because you have some data, this graph is so volatile adding any type of trend is just plain misguiding

Sean :

Wish I could do this in powerbi! Got the beauty this would bring!

// Ven 先生 :

Looks like a bi-weekly x-axis.