Archives 2017

In this blog post, we will see how we can use reverse image search based on (unsupervised) convolutional neural networks to make the analysis of satellite/aerial pictures both more efficient and simpler. After reading this post, you will be able to find similar objects in a large aerial/satellite images and from there develop your own GIS statistical applications (i.e. to count all white cars in your neighborhood, identify specific road markings or kind of trees, etc. ).

The requirement for a (very) large training set is generally the main criticism that is formulated against deeplearning algorithms. In this blog, we show, how deep convolutional neural networks (CNN) can be used in an unsupervised manner to perform efficient reverse image search.

Note:  This blog post was written as a collaboration between Kensu.io and H2O.ai and the blog content was initially posted on on blog.H2O.ia.  You can either read it here, or continue your reading on its original publication page.

In the space of Data Science development in enterprises, two outstanding ...

In this blog, we will finally give an answer to THE question:  R, Python, Scala, Spark, Tensorflow, etc...  What is the best one to answer data science questions?  The question itself is totally absurd, but they are so many people asking it on social network that we find it worth to finally answer the recurrent question using a scientific methodology.  At the end of this blog, you will find a quantitative answer comparing the computing time of each language/library for fitting the exact same Generalized Linear Model (GLM).  Many features matter in the choice of a language/library, among them , the computing and developing time are for sure very important criteria.

Note: This blog post was initially written for the blog of Kensu.io, You can either read it here, or continue your reading on its original publication page.

This blog shows how to perform scalable geospatial data analysis using Geotrellis, Apache Spark, Sparkling-Water and the Spark-Notebook.

As a benchmark for this blog, we use ...

In this blog, we will see how we can perform geospatial data analysis in order to identify new business opportunities.  For this showcase, we will focus on the retail sector and more precisely on the supermarket leading brands in Belgium: Colruyt, Delhaize, Carrefour, and Lidl.  We analyzed the location of supermarkets in Brussels, computed the average time travel to the closest supermarket for Brussels neighborhood and see how these four major brands are sharing their market zone among Brussels neighborhood accordingly.  We are reusing the techniques detailed in the Dynamic Web scrapping blog post.  The techniques described in this post can be useful for all sorts of B2C companies involved in the retail sector, where competition is generally strong and shop implantation matters. ...

In this blog, we will see how we can perform in-depth customer analytics using publicly available inputs from the customers on company Facebook pages.  For this showcase, we will focus on the media sector and more precisely on the RTL group  (leading TV & Radio on the French speaking side of Belgium).  We analyzed the behavior of people acting on the Facebook pages of the RTL group and aggregated all available information to perform per-user analytics and predictions.  We are reusing the techniques detailed in previously published blog posts on Facebook Mining and Sentiment Analysis.  The techniques described in this post can be very useful for all major B2C companies involved in the media, telecoms, retails sectors. ...