New answers tagged

0

People usually test Hurst exponent algorithms using synthetic functions such as (the most common): 1) The random walk. 2) The weierstrass function. 3) ARFIMA In the ARFIMA, you may also introduce some short range dependence. Many Hurst exponent estimators have difficulty to separate between short and long range dependence. Most of this data you can ...


1

Quoting directly from the third paragraph in the introduction of this reference paper, which explains the concept quite intuitively: "We define the embedding dimensionality of a dataset as the number of attributes of the dataset (its address space). The intrinsic dimensionality of a phenomenon (and also of the data retrieved from it) is defined as the real ...


0

Here is a list. I'm sure you'll find something related to healthcare in there. https://www.kaggle.com/datasets https://www.reddit.com/r/datasets/ https://www.gharchive.org/ http://archive.ics.uci.edu/ml/index.php https://dasl.datadescription.com/ https://github.com/awesomedata/awesome-public-datasets Source:https://www.reddit.com/r/statistics/


1

For variables that used in the model (or have potential usage in the model), people usually call them "features" (in machine learning community) or "independent variables" (in statistics community). For columns you mentioned such as name, created at, modified at. We may not use them in the model, but for some reference. We still can call them features, but ...


0

Important takeaways for analysts: Categorizing a continuous predictor will always result in a loss of information "A lot of zeroes" is not 0 inflation, regardless of skew. I can find a negative binomial parametrization that can generate an arbitrary proportion of zeroes and an arbitrarily high maximum. 0 inflation is a theoretical process, like getting ...


0

Should I keep images of my dataset for this purpose or should I use a real life dataset ? Depends on what you're trying to accomplish. If you ultimately care about the performance on your dataset (perhaps you wish to compare with other models also benchmarked on this dataset), then you should evaluate on it. If you ultimately want to deploy the model on ...


0

In general, you should normalize your input using some method. This happens for at least two reasons: 1) Inputs of different magnitudes will affect differently the weights of the neural network. This will make the convergence more difficulty. 2) If the input are larger it can cause the saturation of the intermediate units of the neural network.


1

There are many ways to detect outliers. It is not clear for me if your outliers is in the dependent or independent variable. However, your problem is not about detecting the outlier, since you have already detected. If the outlier is in the dependent variable, maybe one possible way to justify its remotion is to use the cook's distance Cook distance in ...


0

For those interested in the same question: There's a scientific field that handles combination of classifiers. One tries to figure out 1 class with several classifiers, resulting in several estimations, eg, sample 1 classifies as 3,3, and 3, so I define it as 3 based on some criteria. Running your classification over different parts of your data, and then ...


2

The standard procedure is to generate several (e.g. 99) simulations based on the fitted model and then use these to make envelopes of the estimated pair correlation function. This allows you to see the expected variability under the model and then compare with the estimated pair correlation function from your data. This can be interpreted formally as a Monte ...


0

From a purely software perspective, it's a pivot table. Stats-wise, I would call it a cross-tabulation, or a contingency table; it's just not totalled or normalized in any obvious way.


Top 50 recent answers are included