Forecasted Data Set with Analytical Report

PendulumRock Analytics, LLC specializes in the systematic calculation of customized time series data forecasts for sales of different product lines, demand planning, inventory maintenance, room vacancy, passenger availability, financial instruments, and stock prices among any other numerical items that are historically recorded.  Our process is prepared for several data import types including wide format single file data, long format single file data, multiple tab spreadsheet data, multiple file spreadsheet data, Access databases, and SQL databases.  Accommodation can also be made for SQL databases stored through Amazon Web Services.  We take daily, weekly, monthly, quarterly, and yearly data and produce forecasted estimates for the same frequencies.  In addition, the forecasts come with comprehensive analytical documentation.
By filling out a request form, you will provide PendulumRock with the information required in order for us to develop a quote to provide forecasting services for your organization.  In this form, we request the format of your data, the instruments you need to predict, and the periodicity (i.e. daily, weekly, monthly, etc.) you are looking to forecast.  For any other questions regarding details about the request form or report, please contact us.

Why Should You Use Our Service?

PendulumRock saves you money & time in many BIG ways.  Your company will no longer need to go through the following process:

Set up an analytics group

Using your company's Big Data to increase your bottom-line and cut inefficiencies requires a model-building system.  In order to create this architecture, you will need skilled expertise at costly salaries.  Fortunately, we have created this system for you and provide the fruits of our labor at much smaller overhead.  Want to examine the details about what exactly our system produces?

Go through the hiring process

Well, let's say you can afford the overhead costs of expensive Data Scientist salaries and are willing to take that route.  Now, you need to go through the long & tedious process of finding skilled analysts that fit your company's personality and are willing to work as a team.  Opt for our service and you'll be forecasting your horizon while your competitors are sifting through resumes.

Purchase expensive commercial software

Ok, so you can afford the salaries, and you have already assembled your analytics team, and they are ready to get busy building a data-driven system.  Hopefully, they have the ability to program in open-source software.  Otherwise, you are paying out thousands more to a commercial software giant to create a forecasting system.  Save them time and headaches by pointing them to our service.
​Alright, maybe you have developed a worthy team of top-quality analysts and have already made the down-payment on commercial software along with its lengthy & arduous installation, and paid the yearly maintenance fee.  Your team of experts is experienced in programming for this language.  However, there are many different types of time series models and they must sieve through pages of statistics for each model compounded by pages of models, organize the relevant numbers, and then compare across different models.  No current statistical software has effectively automated this process...except us.

PendulumRock provides the most clear & cost-effective path towards prediction analytics

Save your company the stress of the tasks listed above.  Eliminate the noise.

Our service takes into account 4 important principles when selecting the best model to produce your forecasts

Accuracy

After developing a particular model, a wise analyst will back-test it on the data that was used for its creation as well as a sample of that data.  Useful metrics exist to measure how accurately this model predicts, or forecasts, these data sets.  We use a holdout sample in order to determine the best model for predicting each forecast instrument.

Fit

A model can be assessed by determining how much of the variation in the predicted values is actually attributable to the variables used to estimate those predictions.  We use a mean absolute percentage error (MAPE) to determine which statistical model forecasts the data most appropriately by minimizing the MAPE.

Confidence

When establishing variables we think will correctly predict a certain outcome, how assured are we with their ability to do so?  We can derive an amount of certainty by examining the error statistics & model assumptions, which are closely related to accuracy.

Parsimony

When developing models to drive business decisions, how do you know which type of model to choose?  There are so many out there...
In the realm of statistics, it is desirable to choose the model with the greatest accuracy, fit, confidence, and least number of parameters.
Using existing (and expensive) software, analysts must peruse through pages of statistical output to determine which model is the most accurate, has the best fit & confidence, while utilizing the least amount of parameters.  Well, we have solved that equation for you by developing a process that quickly compares all possible time series models and selects the optimal choice within minutes.

Latest Global Shipping News

Latest Global Airline News

Business Process of Model Development & Forecasting