Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Gianluca Malato
Best AI Humanizer Tools in 2025 to Make Your Content Sound Real
By 2025, artificial intelligence will be generating content in almost every field. Whether it’s for a blog, social media, product page or website, artificial intelligence is writing text at very high speeds. Even though this method saves a lot of time, the writing it produces often lacks natural sounding language. That is where AI humanizer tools are useful. With these tools, AI-generated drafts are turned into writing that sounds, flows, and is clear like something written by a person.
Crowd Ignite Review: Affordable Growth with a Strategic Advantage?
There are a lot of risks for all Instagram account owners who want to grow today. First, you can fall into the hands of scammers who steal your data or ask for full access to your account. Second, unscrupulous firms offer followers at an attractive price, but in return, you only grow subscriptions from bots. We have been monitoring Crowd Ignite for a few days. It claims to provide live and engaged subscribers.
An Introduction to the Shapiro-Wilk Test for Normality
Data scientists usually have to check if data is normally distributed. An example is the normality check on the residuals of linear regression in order to correctly use the F-test. One way to do that is through the Shapiro-Wilk test, which is a hypothesis test applied to a sample with a null hypothesis that the sample stems from a normal distribution. Shapiro-Wilk Test Explained Shapiro-Wilk test is a hypothesis test that evaluates whether a data set is normally distributed.
A Practical Introduction to Sequential Feature Selection
Sequential feature selection is a supervised approach to feature selection. It makes use of a supervised model and it can be used to remove useless features from a large dataset or to select useful features by adding them sequentially. The algorithm works according to these steps: Select, from the dataset, the feature that maximizes the average performance of your model in k-fold cross-validation. This dataset is made by only one feature.
A practical introduction to sequential feature selection
Sequential feature selection is a supervised approach to feature selection. It makes use of a supervised model and it can be used to remove useless features from a large dataset or to select useful features by adding them sequentially. The algorithm works according to these steps: Select, from the dataset, the feature that maximizes the average performance of your model in k-fold cross-validation. This dataset is made by only one feature.
A practical introduction to sequential feature selection
12 hours ago With the recent developments in AI and the popularity of ChatGPT, you may want to integrate AI into your marketing practices. Find out how. Technology and artificial intelligence (AI) have infiltrated almost every industry imaginable. For better or worse, AI is here to stay — and it's only getting …
How is your data distributed? A practical introduction to the Kolmogorov-Smirnov test
An introduction to the KS test for beginners Data Scientists often need to assess the proper distribution of their data. We have already seen the Shapiro-Wilk test for normality, but what about non-normal distributions? There’s another test that can help us, which is the Kolmogorov-Smirnov test. Data Scientists usually face the problem of checking the distribution of their data comes.
A practical introduction to the Shapiro-Wilk test for normality
Normality means that a particular sample has been generated from a Gaussian distribution. It doesn’t necessarily have to be a standardized normal distribution (with 0 mean and variance equal to 1).
Does your model beat the baseline?
A baseline model is a model that actually doesn’t use the features, but uses a trivial, constant value for all the predictions. For a regression problem, such a value is often the mean value of the target variable in the training dataset (10 years ago, I used to perform ANOVA tests to compare a linear model with such a trivial model, which was called a null model). For classification tasks, a trivial model just returns the most frequent class in the training dataset.
Is your dataset imbalanced?
Some techniques to see if a dataset is imbalanced Dealing with imbalanced datasets is always hard for a data scientist. Such datasets can create trouble for our machine learning models if we don’t deal with them properly. So, measuring how much our dataset is imbalanced is important before taking the proper precautions. In this article, I suggest some possible techniques. We say that a classification dataset is imbalanced when there are some target classes with very low frequencies than others.
Which models require normalized data?
Normalization is a general term related to the scaling of the variables. Scaling transforms a set of variables into a new set of variables that have the same order of magnitude. It’s usually a linear transformation, so it doesn’t affect the correlation or the predictive power of the features. Why do we need to normalize our data? Because some models are sensitive to the order of magnitude of the features.
Which models are interpretable?
A brief overview of some interpretable machine learning models Model explanation is an essential task in supervised machine learning. Explaining how a model can represent the information is crucial to understanding the dynamics that rule our data. Let’s see some models that are easy to interpret. Data Scientists have the role to extract information from raw data. They aren’t engineers, nor they are software developers. They dig inside data and extract the gold from the mine.
How To Run A/B Tests
Companies often need to compare the results of some action with the results of another one in order to identify the most performing one. Generally speaking, there’s often the need to check which is the best solution among a set of possible solutions. For example, if you run a newsletter, you may want to test different subjects for the same e-mail and then select the subject that gives you the highest open rate. That’s the general idea of A/B tests.
Are your training and test sets comparable?
A short guide on how to create comparable training and test datasets Data scientists usually split a dataset into training and test sets. Their model is trained on the former and then its performance is checked in the latter. But, if these sets are sampled wrongly, model performance may be affected by biases. From the beginning of my career, everybody used to split a dataset into training and test sets randomly and uniformly. It’s a common and generally accepted approach I generally agree with.
A Language-Detection Model in Pure Javascript
How to create Node.js or client-side models that aren’t too CPU-expensive—while still making the most of Python’s libraries Photo by Leonardo Toshiro Okubo on Unsplash Models are very powerful tools to be used in several web applications. The general approach is to make them accessible by using REST APIs. In this article, I’ll talk about a model that can be created in Python and then deployed in pure Javascript.
Visualize the predictive power of a numerical feature in a classification problem
How to visualize the predictive power of a numerical feature Image by author Measuring the predictive power of some feature in a supervised machine learning problem is always a hard task to accomplish. Before using any correlation metrics, it’s important to visualize whether a feature is informative or not. In this article, we’re going to apply data visualization to a classification problem. There are several ways a numerical feature can be predictive against a categorical target.
How to measure outlier probability
A method to calculate the probability that a point is an outlier Photo by Lukas from Pexels Outliers are a great problem for a data scientist. They are “strange points” in a dataset that must be checked in order to verify whether they are errors or real phenomena. There are several algorithms that can help us detect outliers, but we sometimes need to measure the probability that a point is an outlier, rather than a 0–1 value. Let’s see a possible method.
A Mathematical Method for Pricing a Digital Product
A possible use of math to help companies price their products better Photo by Glenn Carstens-Peters on Unsplash Pricing a digital product can be hard, but math can be a nice help to get some useful results. Let’s see how. The main problem of a digital entrepreneur is to select the best price for a product. A high price will reject customers and lower the revenue, as well as a low price which will surely attract new customers, but not enough to have good revenue.
Outlier identification using Interquartile Range
IQR (Interquartile Range) is the difference between thethird and the first quartile of a distribution (or the 75th percentile minus the 25th percentile). It is a measure of how wide our distribution is since this range contains half of the points of the dataset. It’s very useful to make an idea of the shape of the distribution. For example, it is the width of the boxes in the boxplot. Once we calculate it, we can use IQR to identify the outliers.
Increase model stability using Bagging in Python
In machine learning, the prediction error can be decomposed into 3 parts: the squared bias, the variance and the irreducible error. Bias is a measure of how far the predictions are from the real value, variance measures how stable our model is against retraining on a sample created from the original distribution of our dataset and the irreducible error is a noise term that cannot be eliminated. Studying these terms and trying to reduce them is called bias-variance tradeoff.
What managers should expect from Data Scientists
Office Hours Photo by Campaign Creators on Unsplash Data Science has entered the world of big companies, where data is. Managers of such companies often ask things that they don’t actually need and forget to pretend the only useful things to have. “I want an algorithm per month”. Yes, I once heard somebody saying something like that and I’m afraid I’m not alone. How many times did a manager ask you to build algorithms?
A Python library to remove collinearity
Collinearity, often called multicollinearity, is a phenomenon that rises when the features of a dataset show a high correlation with each other. It’s often measured using Pearson’s correlation coefficient. If the correlation matrix shows off-diagonal elements with a high absolute value, we can talk about collinearity.
Don’t start learning data science with neural networks
Data Science is about data, not about models. So, focusing on Deep Learning is like focusing on the models and that is wrong. You must focus on proper Exploratory Data Analysis and Pre-processing before using any model. A good visualization can make us reach good results before using any neural network because our purpose is to extract information from data. So, do yourself a favor and don’t start a machine learning project with neural networks nor any other model.
Hyperparameter tuning: Grid search and random search
Hyperparameters are model parameters whose values are set before training. For example, the number of neurons of a feed-forward neural network is a hyperparameter, because we set it before training. Another example of hyperparameter is the number of trees in a random forest or the penalty intensity of a Lasso regression. They are all numbers that are set before the training phase and their values affect the behavior of the model. Why should we tune the hyperparameters of a model?
How to explain neural networks using SHAP
Image by author Neural networks are fascinating and very efficient tools for data scientists, but they have a very huge flaw: they are unexplainable black boxes. In fact, they don’t give us any information about feature importance. Fortunately, there is a powerful approach we can use to interpret every model, even neural networks. It is the SHAP approach. Let’s see how to use it to explain and interpret a neural network in Python. SHAP stands for SHapley Additive exPlanations.
How to access Google Sheets from Python using Pandas
Image by author Google Sheets is a very powerful (and free) tool for creating spreadsheets. I’ve almost replaced LibreOffice Calc with Sheets, because it’s very comfortable to work with. Sometimes, a data scientist has to pull some data from a Google Sheet into a Python notebook. In this article, I’ll show you how to do it using just Pandas. The first thing to do is to create a Google Sheet. For this example, it will contain just 2 columns, one of which (the Age) has one missing value.
How many neurons for a neural network?
A neural network is a particular model that tries to catch the correlation between the features and the target transforming the dataset according to a layer of neurons. There are several books that have been written around neural networks and it’s not in the scope of this article to give you a complete overview of this kind of model. Let me just say that a neural network is made by some layers of neurons. Each neuron gets some inputs, transforms them and returns an output.
Feature selection in machine learning using Lasso regression
I have already talked about Lasso regression in a previous blog post. Let me summarize the main properties of such a model. It is a linear model that uses this cost function: aj is the coefficient of the j-th feature. The final term is called l1 penalty and α is a hyperparameter that tunes the intensity of this penalty term. The higher the coefficient of a feature, the higher the value of the cost function.
When and how to use power transform in machine learning
Power transfom is a family of functions that transform data using power laws. The idea is to apply a transformation to each feature of our dataset. What’s the purpose of apower transform? The idea is to increase the symmetry of the distribution of the features. If a feature is asymmetric, applying a power transformation will make it more symmetric. Let’s see an example using the breast cancer dataset in scikit-learn.
Engineering Practices for Machine Learning Lifecycle at Google and Microsoft
As demands for AI applications grow, we’ve seen a lot of effort put by companies to build their Machine Learning Engineering (MLE) tools tailored for their needs. There are just so many challenges faced by industries in regards to having a well-designed environment for their Machine Learning (ML) lifecycle: building, deploying, and managing ML models in production. This post will cover two papers, explaining MLE practices from two of the leading tech companies: Google and Microsoft.
How to build your own AutoML library in Python from scratch
An AutoML library is any piece of software that automates some of the hardest (and boring) parts of a machine learning pipeline. Despite doing all these tasks manually, using AutoML will speed the machine learning process avoiding the risk of mistakes.
An efficient language detection model using Naive Bayes
The problem we are going to face is to create a model that, once fed with a text, is able to detect its language. The text can be a sentence, a word, a more complicated text, and so on. The output variable can be, for example, the language code (like “en” for English). A good idea would be to have a model that detects the language of a text even if this text contains words that the model has not seen in the training phase.
The bootstrap. The Swiss army knife of any data scientist
Bootstrap is a technique made in order to measure confidence intervals and/or standard error of an observable that can be calculated on a sample. It relies on the concept of resampling, which is a procedure that, starting from a data sample, simulates a new sample of the same size, considering every original value with replacement. Each value is taken at the same probability of the others (which is 1/N).
An example of overfitting and how to avoid it
Overfitting occurs when your model learns too much from training data and isn’t able to generalize the underlying information. When this happens, the model is able to describe training data very accurately but loses precision on every dataset it has not been trained on. This is completely bad because we want our model to be reasonably good on data that it has never seen before. In machine learning, simplicity is the key.
Visualize multidimensional datasets with MDS
As long as we work with two-dimensional datasets, a simple scatterplot can be quite useful to visualize patterns and events. If we work with three-dimensional data there’s still some chance to visualize something using 3d plots. But what happens if we want to visualize higher-dimensional datasets? Things can become more difficult. Think about clustering problems. It would be very wonderful if we could visualize data in many dimensions in order to check whether there are some patterns or not.
The most used probability distributions in Data Science
The simplest probability distribution is the uniform distribution, which gives the same probability to any points of a set. In its continuous form, a uniform distribution between a and b has this density function: And here’s how it appears. As you can see, the wider the range, the lower the distribution. This is necessary to keep the area equal to one. The uniform distribution is everywhere. Any Monte Carlo simulation starts by generating uniformly distributed pseudo-random numbers.
Why training set should always be smaller than test set
The real task of a supervised model is not learning on the largest dataset possible, but learning in a way that the model performance on unknown data is satisfactory. That’s why we perform the model cross-validation on an unseen dataset. Data scientists usually provide some model performance metrics calculated on such test datasets, like AuROC, accuracy, precision, root mean squared error and so on.
Introducing WikiLeaker - An OSINT Tool for Searching WikiLeaks
WikiLeaker I have been teaching virtual and in-person OSINT courses for about the last 6 months as The OSINTion. When going through the tools, I have always gone out of my way to provide credit where it is due to the tool developers and researchers behind the tools. One tool in particular that stuck out to me is Datasploit. The tool is written by Shubham Mittal and his team at RedHunt Labs. I have been an adherent of the tool since I learned that OSINT was, well…OSINT (h/t to Justin Seitz for that).
How to correctly select a sample from a huge dataset in machine learning
Let’s make a simple mental experiment. Imagine that we are in a library and want to learn Dante Alighieri’s Divina Commedia word by word. We have two options: Grab the first edition we find and start studying from it Grab as many editions as possible and study from all of them The right answer is very clear. Why would we want to study from different books when just one of them is enough? Machine learning is the same thing.
Covid-19 infection in Italy. Mathematical models and predictions
Every day, the Italian Civil Protection Department refreshes the cumulative data of infected people. This data is publicly available as open data on GitHub here: https://raw.githubusercontent.com/pcm-dpc/COVID-19/master/dati-andamento-nazionale/dpc-covid19-ita-andamento-nazionale.csv My goal is to create models of the time series of the total number of infected people to date (i.e. the actually infected people plus the people who have had been infected).
What is Data Science?
By Gianluca Malato, Data Scientist, fiction author, and software developer. Photo by Luke Chesser on Unsplash. Data Science is considered as one of the most modern and fascinating jobs of our time. It can be funny and can give you satisfaction, but is it really as it’s described? In this article, I’ll show you the reality of a Data Scientist’s life.
Feature selection by random search in Python
By Gianluca Malato, Data Scientist, fiction author, and software developer.. Feature selection has always been a great task in machine learning. According to my experience, I can surely say that feature selection is much more important than model selection itself. Feature selection and collinearity I have already written an article about feature selection.
Why Artificial Intelligence is not a technological revolution
Many can say that yes, AI is a theoretical tool made much more real by those nerds at Google, but the reality is the opposite: Artificial Intelligence is not a tool at all. It’s more like a synthetic partner in our lives. It’s something able to use cognitive capabilities in order to perform certain tasks faster than we can. So it’s not a tool; it’s an artificial extension of our brain. Basically, we have already surrogated part of our intelligence to the machine. Let’s think about memory.
How to Correctly Select a Sample From a Huge Dataset in Machine Learning
Photo by Lukas from Pexels In machine learning, we often need to train a model with a very large dataset of thousands or even millions of records. The higher the size of a dataset, the higher its statistical significance and the information it carries, but we rarely ask ourselves: is such a huge dataset really useful? Or we could reach a satisfying result with a smaller, much more manageable one?
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Gianluca
Contact Gianluca, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack