Human genes names are mistaken as dates, limits on the maximum number of rows simply delete additional rows without warning the user, and it also happened that someone lost around 6B$ for a spreadsheet error.
What it’s slowly coming out is that when dealing with the so-called big data, AKA huge collections of data, usually containing more than 10K rows, spreadsheets are not the tool to go.
There are several limitations in many aspects, like the maximum number of rows and columns, auto-formatting which can lead to mistakes and, of course, since spreadsheet is extremely widespread, different users having different literacies uses them, creating unpredictables situations, especially when a situation escalates and the tool is the only tool used. …
A couple of weeks ago, a friend of mine, sent me a link to an extremely interesting video regarding what I have re-baptized: The Startup challenge.
Pieter Levels explains the journey that brought him to bootstrap Nomadlist and while doing so, he talks about a framework he invented and tested on himself: the 12 startups in 12 months challenge.
Pieter, the founder at Nomadlist, claims that developers, musicians, creative peoples have a few problems in common: finishing things.
When our projects are close to the finish, we forget about them and go to the next one, without even launching them. …
In this article, I want to present some findings in the field of social media mining, describing the implementation of a Word2Vec model applied for indexing an entire user base, providing a tool to find similar users within a community.
Although many different social media platforms already offer ways to discover similar user, this set of features are mainly built for the final user, meaning that the goal is to actually display them what they want to see and not users that are actually similar to them under a business point of view.
Algorithms able to target similar users are used behind tools such as Facebook Ads, which gives the advertiser the possibility to target users that are similar to a specific set of conditions such as brands, tastes, or other demographics data. …
In this guide, I want to show you how to make time-series predictions of revenues based on real-life retail data, for these tasks I will be using a very common library: Prophet, developed by scientists at Facebook.
According to Prophet GitHub page:
“A tool for producing high-quality forecasts for time series data that has multiple seasonality with linear or non-linear growth”
Moreover, Prophet is integrated into the AWS ecosystem, making it one of the most commonly used libraries for time series analysis.
In October 2019 started my journey as a data scientist. Here I want to share my experience, tips, and things I wished I knew before to avoid you getting burned.
At the beginning of October, I will turn one year working as a data scientist at Evo. Creating awesome tools with cutting-edge technologies applied to the retail and supply chain industry.
This year has been an incredible rollercoaster, I have learned a lot of things, and I consider myself very lucky to be part of an awesome team, which strives for perfection while being helpful, motivating, and competent.
You are the Average of the 5 People You Spend the Most Time With. …
Statistical modelling gives you the ability to asses, understand and make predictions about data, it is at the very bottom of inferential statistics and can be considered of those “must know” topics.
In statistical analysis, one of the possible analyses that can be conducted is to verify that the data fits a specific distribution, in other words, that the data “matches” a specific theoretical model.
This kind of analysis is called distribution fitting and consists of finding an interpolating mathematical function that represents the observed phenomenon. …
The idea to write this article happened when I had to personally tackle this issue and I did not find any helpful resource.
This article is short and is meant to be short. Lambda functions are made to make you save time, so this article is.
Sometimes tutorials are too long, sometimes they just show a particular use case which you don’t care about, something it’s just written so badly, however, after the need to create a microservice triggered by an API to process some data the struggle started. But chill, I have an answer!
First, you need a
In this tutorial, I am showing you how to use Reinforcement Learning to automatize an autonomous warehouse robot to find the optimal path between different locations.
The use of robotics is constantly expanding in every business sector. Automation takes repetitive tasks and aims to remove the manual input in order to optimize processes and cut costs.
In 2012 Amazon purchased Kiva Systems, a company that develops warehouse robots and related technologies, Kiva was acquired for $775 million. Moreover, many other companies, like Alibaba, Volkswagen, or Geek+ constantly implement robots and related technologies.
Getting started with such topics can be a struggle for a beginner, that is why I believe it is important to put things into context and then to start diving into details. …
Removing a person from a background is indeed an interesting and awkwardly surprising task. In this guide, I want to show step by step how to remove a person from a live stream using OpenCV with Python.
A few days ago, while scrolling hacker news I was impressed by this project made by Jason Mayes, an engineer at Google. Since that moment I first saw that project I decided I must (at least attempt) to replicate it. However, since my to-go tools are mostly R and Python I decided to implement it in the latter.
Among many different techniques for object detection, Facebook came up with its model: Detectron2. This model, similarly to Yolo models, is able to draw bounding boxes around objects and inference with a panoptic segmentation model, in other words, instead of drawing a box around an object it “wraps” the object bounding its real borders (Think of it as the smart snipping tool from photoshop.)
The purpose of this guide is to show how to easily implement a pretrained Detectron2 model, able to recognize objects represented by the classes from the COCO (Common Object in COntext) dataset. This guide is meant to provide a starting point for a beginner in computer vision, it aims at explaining what are the first steps to implement a pre-trained model, and its final goal is to spike your interest into learning more, and arranging your thoughts in this overwhelming field. …