Writing predictive web services with Azure MLValentin Bazarevsky

Why?• According to a recent Gartner report, the business intelligence market is

growing nine percent per year, will exceed $80 billion by 2014, with about 50 percent from predictive analytics by that time. Despite all this, the best opportunity for you is still the one you love and know the best, that no one else has recognized. The possibilities are endless, so why haven’t you started yet?

Tasks• Targeted direct marketing.

• Predictive advertisement targeting.

• Fraud detection.

• Investment risk management.

• Customer retention with churn modeling.

• Movie recommendations.

• Education – guided studying for targeted learning.

• Political campaigning with voter persuasion modeling.

• Clinical decision support systems.

• Insurance and mortgage underwriting.

Analogues• Weka

• Orange

• R

• Scikit-learn

• Tons of them

Typical flow (scikit-learn way)

Typical flow (Dlib way)

Orange

Weka

Azure ML

Environments

One solution

One infrastructure

Bla-bla-bla

Pricing

Forecast use of a city bikeshare systemBike sharing systems are a means of renting bicycles where the process of obtaining membership, rental, and bike return is automated via a network of kiosk locations throughout a city. Using these systems, people are able rent a bike from a one location and return it to a different place on an as-needed basis. Currently, there are over 500 bike-sharing programs around the world.

The data generated by these systems makes them attractive for researchers because the duration of travel, departure location, arrival location, and time elapsed is explicitly recorded. Bike sharing systems therefore function as a sensor network, which can be used for studying mobility in a city. In this competition, participants are asked to combine historical usage patterns with weather data in order to forecast bike rental demand in the Capital Bikeshareprogram in Washington, D.C.

https://www.kaggle.com/c/bike-sharing-demand

Data datetime - hourly date + timestamp season - 1 = spring, 2 = summer, 3 = fall, 4 = winterholiday - whether the day is considered a holidayworkingday - whether the day is neither a weekend nor holidayweather - 1: Clear, Few clouds, Partly cloudy, Partly cloudy2: Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist3: Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds4: Heavy Rain + Ice Pallets + Thunderstorm + Mist, Snow + Fogtemp - temperature in Celsiusatemp - "feels like" temperature in Celsiushumidity - relative humiditywindspeed - wind speedcasual - number of non-registered user rentals initiatedregistered - number of registered user rentals initiatedcount - number of total rentals

Findings• Still just a beta.

• Excellent for education purposes.

• Low psychological barrier.

• Easy to start and deploy.

• No continuous re-learning.

• Only simple use-cases

Links• http://azure.microsoft.com/en-us/documentation/articles/machine-learning-

faq/

• http://azure.microsoft.com/en-us/documentation/articles/machine-learning-azure-ml-netsharp-reference-guide/