Sign in

A consulting firm, top expert in digital/data science from other leading management consulting firms like Accenture. Guest author on top publications.

Traditionally, B2B marketers often try to go after an entire industry, while ABM focuses marketing resources on a smaller set of high-value accounts, and fully leverages the breadth of digital channels to augment offline sales efforts.

However, many companies failed when trying to implement ABM. One of the reasons is believed to be that ABM is usually marketing-led and lacks alignment across the organization. Marketers started to shift to orchestrating plays across departments and emphasizes the coordination needed to deliver highly integrated customer experiences.

Then, can we measure how different marketing activities and engagements are interacting with each other? The…


Decision trees, bagging, random forest and boosting can all be applied to both regression and classification. Decision trees are simple to understand by people who aren’t comfortable with mathematics, but typically not competitive with the best supervised learning approaches in terms of prediction accuracy. Bagging, random forests, and boosting grow multiple trees which are then combined to yield a single consensus prediction, which often results in dramatic improvements in prediction accuracy.

Decision tree

If the decision tree is too large, overfitting may happen. An extreme example could be one observation each terminal node or leaf.

However, what if we stop…


How to evaluate a neural network model? We will break this topic into two parts. Part I and Part II.

K-fold cross validation

We usually split data into training, validation (development), and test sets. We train different models initialized with different hyperparameters, compare their performance on the validation set, and then pick the hyperparameters associated with the best performance to use in our final model, which will be evaluated on the holdout test set. K-fold cross validation is similar, but instead of just creating one split, we split the datasets into K subsets (folds) and then train on K-1 of…


Most of the time, underfitting happens when the models or algorithms are too simple to fit more complex trends. The solution to this is simply train your model for more time or try a more complex model or algorithm. So, what is overfitting?

What is overfitting?

We briefly talked about overfitting when discussing methods for high-dimensional problems in linear regression. Overfitting is when the model fits the idiosyncrasies of the training data patterns so well that the model will only work well for the training data, and not on new data sets. Overfitting happens when the model fits to not…


Most of the relationships are not linear in the real world. However, linear model is an important model. It has distinct advantages in terms of its interpretability and predicts future data quite well. Also, if you use an over-complex model, like a very deep neural network that can accommodate a high number of feature weights and that uses a high number of predictor variables, to solve a simple linear model problem especially in small datasets, you could experience overfitting, which is when the model fits the training data so well that it does not generalize well on other datasets.

Today…


A data management platform (DMP) is the backbone of data-driven marketing and unifies and organizes your first-, second-, and third-party data.

What is a DMP? A DMP, data management platform, is the backbone of data-driven marketing and serves as a unifying platform to collect and organize your first-, second-, and third-party data.

It holds data about prospects and customers from different online and offline sources and activates them through multiple channels such as display and search.

DMPs are key to programmatic advertising — but you will often meet folks in this space who are not aware of certain features that…


The pandemic presents an opportunity and a need for many companies to build the competences they wish they’d invested in before: to be more digital, and data-driven. To B2B marketers, account-based attribution becomes more relevant than ever.

What is account-based attribution?

According to Gartner, the typical buying group for a complex B2B solution involves six to ten people. Account-based attribution is a technique for grouping stakeholders into accounts (companies) and relating a desired business outcome in the form of “credit” to customer interactions so organizations can understand better how the marketing and sales activities are creating revenue. …


Missing data are often a problem in statistical modeling. They arise frequently in practice and are caused by many circumstances. For example, study subjects might fail to answer questions on a questionnaire, data can be lost, a survey might not receive enough responses, and so on.

Here are some of the techniques to treat missing value -

  1. Ignore the records with missing values.

Many tools ignore records with missing values. When the percentage of records with missing values is small, we could ignore those records.

2. Substitute a value such as mean, median.

When the percentage is large and also when it makes sense to do something to avoid bias modeling…


B2B Digital experience has changed in the last 6–7 months. Long before Covid-19, most companies saw the opportunity to engage their B2B customers more effectively online. Since the Covid-19 crisis began, to be more digital and data-driven have become top priorities.

B2B sales leaders have moved from being “forced” to adopt digital in reaction to the widespread shutdowns in the early stages of COVID-19 to a growing conviction that digital is the way to go, according to a survey of B2B businesses conducted by McKinsey.

Here are a few things organizations can do to accelerate analytics and AI in order…


How do you deepen your relationships with customers at scale? What’s going on with the customers who didn’t complete their purchases and never responded to your surveys. How do you transmit these customer insights to frontline employees who deliver personal and attentive service? Analytics and machine learning can come in to help. Here are some use cases.

Descriptive

Most companies have rule-based segmentation in place to get initial insights. It is part of the first stage of analytics which answers the question what happened. More and more companies integrate not only inputs from demographic-based data but also consider a wide…

AlphaConverge

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store