Developers Club geek daily blog

AI, BigData &HPC Digest #1 (on May 7? On June 3)

3 years, 5 months ago
The FlyElephant team has prepared the latest issue of the digest which includes selection with links to news and materials in the directions: artificial intelligence, big data and high-performance calculations. Pleasant reading!

Read more »

Deep learning on R, we train word2vec

3 years, 5 months ago
Word2vec is almost only algorithm of deep learning which can rather easily be started on the normal PC (but not on video cards) and which builds the distributed submission of words for acceptable time, at least so consider on about here what focuses can be done with the trained model, I have understood that such piece is simply obliged to try. Problem only one, I mainly work at language R, and here official implementation of word2vec under R to me did not manage to be found, I think it simply is not present.

Read more »

Recognition of physical activity of users with examples on R

3 years, 6 months ago
The problem of recognition of physical activity of users (Human activity Recognition or HAR) came across to me only as educational tasks earlier. Having opened for itself possibilities of Caret R Package, convenient wrapper for more than 100 algorithms of machine learning, I have decided to try it and for HAR. In UCI Machine Learning Repository there are some data sets for such experiments. As the subject for me is not really close to dumbbells, I have selected recognition of activity of users of smartphones.

Read more »

AI, BigData &HPC digest #0. Pilot release

3 years, 6 months ago
Hi, Habr!
My name is Dmitry Spodarets. I am the founder of the FlyElephant project who provides to research associates and engineers the environment for execution of computing programs. In our team we actively monitor three directions: artificial intelligence, big data and high-performance calculations. Interesting materials with which we will acquaint you within this digest constantly come across to us.

Read more »

Viewing of configuration of arrays of Storwize with R? Shiny Dashboard

3 years, 6 months ago
Storwize disk arrays of the IBM company are well-known in the market, and here with convenient preview player of their configuration (especially, without having access to the array), except for the Total Productivity Center software product from IBM, already anything and is not present (svcmon scripts are not supported by perl any more). And employees and, or the partners analyzing configurations of these arrays often have such requirement. Therefore, I propose the solution on viewing of XML files of configuration of these arrays.

In this article I will describe how it is possible to read xml information to present it in the form of tables, summary information on deshborda, and all this in the form of web application. As appears from the name of article? it is implemented on R, with packet (framework) for web applications to R? Shiny dashboard.


Read more »

R as the instrument of monitoring of the prices

3 years, 7 months ago
In this article would like to concern such subject as monitoring of competitors. I understand that at this subject is as many supporters, after all anyway monitoring is necessary for successful development of almost any company, and opponents who protect interests of the business from monitorshchik.

Those who as that is connected with sales in the competitive market, for certain know that monitoring of competitors is important task. Results are used for absolutely different purposes? from change of local pricing policies and maintaining the range before drawing up strategic development plans of the company. The author has decided to practise solution of this task and to promonitorit one of large retailers of electronics in Russia, whose regular client the author is. What from this left?

Read more »

Bag of words and the sentiment-analysis on R

3 years, 7 months ago
This article is prepared based on (the first part) the educational Bag of Words Kaggle task, but it not transfer. The original task is made on Python. I wanted to evaluate possibilities of language R for processing of texts in natural language and at the same time to try implementation of Random Forest in R-packet caret wrapper.

Sense of task – to construct "machine" which will process definitely reviews of movies in English and to define overview tonality, carrying it to one of two classes: the negative/positive. As the training selection in task the data set with twenty five thousand revues from IMDB marked by unknown volunteers is used.

Read more »

Visualization of statistics of productivity of the equipment with R – Shiny

3 years, 8 months ago
"In the 21st century not those will be illiterate,
who is not able to read and write,
and those who is not able to study,
to forget and be retrained"

Alvin Toffler

IT specialists can have tasks connected with performance review of the equipment or the analysis of results of different load generators (ioMeter, Vdbench and other). In most cases for these purposes Excel with creation of time series is used, with finding of the main descriptive statistics and attempts everything somehow to analyze it. There is alternative means of faster and convenient analysis of descriptive statistics with various charts and possibility of creation of web application for the general access. I will not concern the real statistics with different methods of data analysis, only basic descriptive statistics (without verification of tests and even p-value will not be) and different charts.

In this article I will describe one of options of how it is possible to analyze such information, to present it in the form of charts (traffic!), and all this in the form of web application. As appears from the name of article – it is implemented on R, with packet (framework) for web applications to R – Shiny.

Read more »

Application of machine learning in trading

4 years, 2 months ago
Note of the translator 1. I have come across this blog in one of reviews of materials on machine learning. If you well understand machine learning, you will not find in this article for yourself anything interesting. She rather surface also mentions only bases. If you, as well as I, only start being interested in this subject, welcome under kat.
Note of the translator 2. Code will be a little, and that that is is written in language R, but you should not despair if you still never in eyes saw it. To this article I too did not know anything about it therefore I have specially separately written? spur? on language, having included everything there that to you will occur in article. If you wish to understand, I recommend to begin with small course on interesting information and useful links. And at last there is big crib.
Note of the translator 3. Article from two parts, however the most interesting begins only in the second part therefore I have dared to integrate them in one article.

Part 1

In this series of articles, I am going to construct and ottestirovat step by step the simple strategy of management of asset based on machine learning. The first part will be devoted to basic concepts of machine learning and their application to the financial markets.

Machine learning is one of the most promising directions in financial mathematics, in recent years received reputation of the sophisticated and difficult tool. Actually everything is not so difficult.

Read more »

R + C + CUDA =?

4 years, 6 months ago
Sometimes there is need to accelerate calculations, and it is desirable at once many times. Thus it is necessary to refuse convenient, but slow tools and to resort to something more low-level and fast. R has quite developed opportunities for work with the dynamic bibioteka written on With/with ++, Fortran or even Java. I on habit prefer With/with ++.

Read more »