Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 12 Next »

Discretized Attributes 

Discretized attribute is a function attribute which allows its discretization based on a set of “less-than” thresholds. For a given value, the method will associate the symbol of the first threshold. If none of the thresholds matches the value is associated to the default symbol. 

To create a discretized attribute:

  1. Click Transform > Discretized attribute in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new discretized attribute, default prefix: discretized_
  4. Select Input attribute.
  5. Enter Default symbol, example: OK.
  6. Click Add to select the appropriate Thresholds (>=).
  7. Enter the Value and the Symbol, example: Threshold (>=) 15 NOK.
  8. Click Compute.

Discretized Time Attributes

Discretized time attribute is a function attribute which allows to extracting relevant and understandable information of a time attribute.

To create a discretized time attribute:

  1. Click Transform > Discretized time attribute in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new discretized time attribute, default prefix: DATE_
  4. Select time attribute.
  5. Enter Temporal unit, selection among: None, Unix time (s), Unix time (ms) and Excel time.
  6. Enter Locale, which influences the language and format of the symbolic output attribute (month, day), selection among: English (en-US) and French (fr-FR). 
  7. Enter Time zone, see the following: https://goo.gl/QiYgye.
  8. Select the output formats.
  9. Click Compute.

The available outputs are:

  • Year: numeric (format: YYYY)
  • Quarter: symbolic (format: “Qn”, with the quarter number)
  • Month: numeric (format: MM)
  • Day of the week: numeric (format: D)
  • Day of the week: symbolic (format: “name of the day”)
  • Year-month: symbolic (format: ”YYYY-MM”)
  • Year-quarter: symbolic (format: “YYYY-Qn”)
  • Second of the year: numeric (format: a number )
  • Second of the day: numeric (format: a number )

Fill Missing Values

Fill missing value creates an attribute or set of attributes filling the missing values with a value designated by the strategy adopted, this missing value can be replaced by a default value, can be an average value of previous and next objects, can be the previous or the next values or an interpolation of the previous and next objects. 

To create a strategy of fill missing value of a certain attribute:

  1. Click Transform > Fill missing values in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new attribute set.  
  4. Select object set.
  5. Enter the attributes or the set of attributes.
  6. Enter Index, if required.   
  7. Choose the Type.
  8. Enter a Value, in case, Type chosen is Default Value.
  9. Enter the attribute name, default prefix: CLEAN_
  10. Click Compute.

The types used:

  • Average: the arithmetic mean of the value of the previous object and the value of the next object  
  • Interpolation: the linear interpolation between the value of the previous object and the value of the next object  
  • Previous: the value of the previous object
  • Next: the value of the next object  

Shifted Attributes

Shifted attributes creates, for each attributes, an object containing values within a defined time window. The window is specified with a lower and upper bounds and, a step.

To create a discretized attribute:

  1. Click Transform > Shifted attribute in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new attribute set.
  4. Select Input attributes or attribute set. 
  5. Enter Lower shift, negative integer value for backward index, default value: - 10.
  6. Enter Upper shift, positive integer value for forward index, default value: 0.
  7. Enter Step, integer positive value, default: 1.
  8. Click Compute.

 

Moving Averages

Moving average is a calculation to analyze data points by creating a series of averages of different subsets of a data set. A common practice is to use moving average to smooth noisy data. 

To create moving average of a set of attributes:

  1. Click Transform > Moving averages in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new attribute set.
  4. Enter prefix, default: MOV_AVG_
  5. Select Input attributes or attribute set. 
  6. Enter Window left, positive integer value, default value: 0.
  7. Enter Window right, positive integer value, default value: 0.
  8. Click Compute.

 

Differentiated Attributes

Differentiated attribute creates derivative of attributes. The derivatives are computed for each input attribute. 

To create a differentiated set of attributes:

  1. Click Transform > Differentiated attributes in the menu.
  2. Change the data source, if required.
  3. Enter a name for the new attribute set.
  4. Enter prefix, default: DIFF_
  5. Select Input attributes or attribute set. 
  6. Enter Step duration.  
  7. Enter Differential method.
  8. Click Compute.

The derivative options are:

  •  Newton: difference quotient (order 1)
  • Symmetric: difference quotient (order 2)
  • High order: five-point stencil 
  • No labels