NumXL is a suite of lightweight time series Excel add-ins. It transforms your Microsoft®1 Excel® application into a first-class time series software and econometrics tool, offering the kind of statistical accuracy offered by the leading statistical packages. NumXL integrates natively with Excel, adding scores of econometric functions, a rich set of shortcuts, and intuitive user interfaces to guide you through the entire process.
Whether you have a simple homework problem or a large-scale business project, NumXL simplifies your efforts. It helps you reach your goal in the quickest, most thorough way possible.
NumXL keeps your data and results connected in Excel, letting you trace your calculations, add new data points or update an existing analysis, easily sharing your result with co-workers - and, yes, even with your boss.
The learning curve couldn't be easier: NumXL requires no programming or scripting. You will not have to move your data between any external programs.
You can also do any kind of ad-hoc analysis, as all of NumXL functions are accessible in your spreadsheet, and inside VBAi environment should you choose to script.
What can NumXL do for me?
NumXL comes packed with scores of functions that you can easily access through the function wizard in Excel, as well as several wizards and shortcut UIs to facilitate the time series analysis process and automate the most common steps (e.g. summary stats, modeling, calibration, diagnosis, forecast, and more.)
1. General Statistics
Using the descriptive statistics and correlogram wizards, you can examine the data series summary and time series statistics with just a few clicks. The wizards come with an extensive set of statistical tests, from a simple one-sample test for mean to the more sophisticated normality and ARCHi effect tests.
Furthermore, examining time dependency (auto correlation) in your sample data is only a few clicks away.
The wizards generate professionally organized tables and graphs summarizing all your calculations, ready to be included in your presentation. To make things even easier, all outputs include NumXL functions in their formulae for connecting values with inputs, so you can edit, update or customize as you see fit. You can even re-run the wizard if you feel lazy.
2. Correlogram Analysis
Using the correlogram wizard and function, constructing autocorrelation and partial autocorrelation plots is a snap.
Furthermore, NumXL comes with support for calculating cross correlation using three different methods: Pearson, Spearman and Kendall.
Finally, for completion purposes, NumXL support Hurst exponent analysis and GINI coefficient.
3. Statistical Testing Demo Video
For serious data analysis, you might need to consider the statistical significance of a calculated parameter (e.g. autocorrelation factor, excess kurtosis, etc.) or verify an assumption from earlier data (e.g. Normality, stationarity, absence of serial correlation and others).
NumXL packages the majority of these tests as a simple API to calculate the P-Value of the underlying test.
Data transformation is a common preliminary step in real-world analysis and/or modeling. NumXL comes with most common transformation functions (e.g. Box-Coxi, difference and seasonal difference/integral operators, and others.)
5. Smoothing Demo Video
Smoothing and filtering are two of the most commonly used time series techniques for removing noise from the underlying data to help reveal the important features and components (e.g. trend, seasonality, etc.). However, we can also use smoothing to fill in missing values and/or conduct a forecast.
NumXL supports several smoothing functions, from a simple weighted-moving average (WMAi) to Winter’s triple exponential smoothing function.
Of course, we can’t talk about smoothing without mentioning trend functions. Trend analysis is very often used (or abused) in the industry to make a quick (and dirty) forecast. Executives might use the trending tool as a sanity check when he/she examines results from more advanced models. NumXL supports several forms of trend: linear, polynomial, power, exponential and logarithmic.
6. Calendar functionality Demo Video
Calendar events influence the values of the time series sample, and a prior adjustment for those events will help us to better understand the process, modeling and forecast.
NumXL comes with numerous functions to support calendar adjustment, date rolling and adjustment, U.S. and non-U.S based holidays support, non-western weekends, and public and bank holiday calendars.
7. Spectral Analysis Demo Video
In statistics, spectral analysis is a procedure that decomposes a time series into a spectrum of cycles of different lengths. Spectral analysis is also known as frequency domain analysis.
NumXL currently support the discrete Fourier transformation (and its inverse), with future plans for more extensive coverage.
8. ARMA/ARIMAi Demo Video
NumXL comes loaded with numerous functions to help you with any ARMA analysis task. You can start by specifying the model’s order using the ARMA wizard. The wizard will tie in model-related calculations – the Log-likelihood function, Akaike’s information criterion (AICi) and residual diagnosis - with your input data. Once that is done, fitting the model parameters (calibration) is a snap: simply select the model and click on "calibration." The same goes for forecasting: select the model table and click "forecast".
You can always edit the formulas in the output cells, get intermediate calculations (e.g. residuals, fitted mean) or use a NumXL function in your own formulas or VBA code if you so choose.
9. Seasonal ARIMA Demo Video
NumXL support seasonal ARIMA through two different models: (1) AirLine, and (2) X-12-ARIMAi.
X-12-ARIMA is the widely used seasonal adjustment program developed, supported and maintained by U.S. Census bureau. NumXL provides an intuitive interface with the program to help Excel users make forecasts and seasonal adjustments quickly and efficiently for economic and financial data.
NumXL also offers users access to all raw files (input/output) that are generated in the process of the data analysis.
10. ARCH/GARCHi Demo Video
Similar to ARMA/ARIMA, modeling a GARCH-type model is a breeze. Using the GARCH Wizard, you can generate a model output table with all coefficient values and related calculations (e.g. LLFi and residual diagnosis). This table can be used to calibrate the model and predict out-of-sample values.
Furthermore, NumXL supports Gaussian, Student's t and GEDi-type innovations.
11. ARMA-GARCH mixture model
In the event you wish to model the time-varying conditional mean and the conditional volatility in one model, an ARMA-GARCH mixture model is in order. By combining the two models, the ARMA will follow the mean and pass the residuals to GARCH to follow the variance over time.
ARMA-GARCH combo models support all the ARMA and GARCH models supported solo, including the non-Gaussian innovations.
12. Generalized Linear Model (GLMi) Demo Video
Whether you have a logistics regression or a general linear model in mind, you can use the generalized linear model (GLM). NumXL uses the GLM wizard to help you specify the input data, link functions (e.g. probit, logit, log complement) and generate a model output table.
As with all of our models, the output model table is used to calibrate and perform out-of-sample forecasts.
13. Miscellaneous Demo Video
Throughout our analysis, we implemented a few functions that did not fit in with any of those listed above: interpolation, extrapolation, least squares regression and regular expressions functions. We combined these orphan functions into a single category called “Utilities”.
- 1. Microsoft and the Office logo are trademarks or registered trademarks of Microsoft Corporation in the United States and/or other countries