Statistical Training and Seminars

We offer cost-free technical support (up to 2 hours) to participants who need help with implementing the proprietary techniques reviewed during the workshops. Most of our techniques can be implemented at very low cost, without the need to purchase expensive statistical software. Even advanced concepts such as ridge regression are explained in simple English. Software engineers or product managers are welcome to attend. Source code and/or detailed algorithms are delivered to all participants. The number of participants is limited.

• Web Analytics, Click Scoring and Click Fraud Detection

  Please check our web anlytics seminar webpage for further details about pricing and upcoming events. Topics usually include state-of-the art patent pending technology for text mining, web mining, query intelligence, click scoring or fraud detection.

  One day

• Effective Advertising for Small Businesses

  After completing our training, you will be able to set up effective pay-per-click advertising campaigns with Google and Yahoo, and identify other potential high ROI publishers for your advertising needs. You will learn how to avoid click fraud and other pitfalls. You will also learn how to find and purchase keywords efficiently and benefit from other technologies that will provide you with a competitive advantage: automated word of mouth advertising and automated reverse word of mouth advertising. We also discuss other topics such as

  • How to get the best out of competitive intelligence?
  • Writing business proposals that sell
  • Strategic alliances
  • Identify unfair competitive advantages to beat your competitors

  Check our advertising seminar webpage for details.

  Two hours, $195.

• Computational Finance - Stock Market Data Mining

  Automated downloads of historical stock price data. State-of-the-art pattern recognition for stock market data. Unusual patterns. Designing robust trading signals. Creating a black box with automated trades. Participants receive proprietary software with source code to process several hundred stocks and design efficient strategies with daily signals. Check our stock market seminar webpage for details.

  One day, $795.

• Business Statistics

  Robust multivariate regression with a proprietary algorithm, review of non linear techniques including logistic regression, clustering, density estimation and decision trees. Time series, censored data and detection of periodicity. Monte Carlo simulations to perform sophisticated statistical tests and compute confidence intervals for regression parameters. Application to extreme values and quality control (A/B testing). Pareto analysis. Re-sampling to perform robust one way or two way ANOVA with non Gaussian data. Model selection, data reduction, selection of variables, predictive modeling, robust statistics, outlier detection. Applications to real life marketing, finance and risk management problems. Participants receive proprietary software with source code for regression and Monte Carlo simulation.

  One day, $595.

• New Regression Techniques

  This course focuses exclusively on the state-of-the-art regression techniques developed by Data Shaping Solutions. Participants receive proprietary software with source code for regression and Monte Carlo simulation.

  Four hours, $395.

• Advanced Statistical Graphics

  Introduction to R and S-Plus. Palettes. Zip code data. Maps. Spatio-temporal data. Unlocking the meaning of complex multivariate data. Efficiently displaying several variables using palettes, color codes, fonts and symbols. Participants receive source code and zip code database to generate high quality maps.

  Four hours, $395.

• Data Mining Very Large Datasets

  Efficient Ad-Hoc Techniques. Data reduction. Model selection and evaluation. Deriving powerful metrics from a set of initial categorical variables. Experimental design to build robust decision trees. Monte Carlo simulations and extreme values. Pareto analysis. Logistic regression applied to rank statistics and bin counts. Non linear discriminant analysis. Pattern recognition. Markov processes and survival time. Fast algorithms: design and optimization. Comparing SAS, Perl and C to process very large datasets. Efficient SAS code for large datasets. Applications: credit card fraud detection, stock trading systems, image analysis.

  One day, $595.

• Statistical Methodology

  Review of the protocol used to perform a statistical analysis and deliver useful and reliable results:

  • Writing a proposal
  • Writing a requirement document or specifications
  • Identifying metrics of interest
  • Gathering, cleaning and merging datasets
  • Exploratory analysis, data reduction
  • Model selection, model testing, model fitting
  • Calibration
  • Model evaluation, sensitivity analysis
  • Presentation of results to management
  • Maintenance

  We will use an advanced regression analysis to illustrate the process. Participants receive proprietary software with source code for regression and Monte Carlo simulation.

  Four hours, $395.

• Advanced Techniques for Decision Trees

  Tree boosting and tree forests. Algorithms to efficiently process thousands of features when dealing with very large data sets. Regression trees. Trees with embedded logistic regression. Tree growing.

  Four hours, $595.