Data analytics on measurement data
Disclosed are various example embodiments which may be configured to: obtain, for each of one or more network entities, a time series of values of a measurement parameter for respective timestamps, the time series including measured values and a special numerical value at one or more timestamps; parse the time series of values to determine which numerical value in the time series of values corresponds to the special numerical value; assign flags to the values in the time series of values, wherein a flag assigned to a value obtained at a given timestamp indicates whether the measurement was or not available at the given timestamp in the time series of values.
1. A method, comprising:
obtaining, for each of one or more network entities, a time series of values of a measurement parameter for respective timestamps, the time series of values including measured values and a special numerical value at one or more timestamps, the time series of values including the special numerical value at a given timestamp for replacing a value of the measurement parameter when no measured value is available for the measurement parameter at the given timestamp;
parsing the time series of values to determine which numerical value in the time series of values corresponds to the special numerical value, the parsing including
detecting same-values sequences having a minimum length in the time series of values,
generating a set of at least one value including the value of each of the detected same-values sequences having the minimum length, and if the set of at least one value includes a sole value, identifying the sole value as the special numerical value, and
computing a count of value changes occurring in a sliding time window of a given length applied to the time series of values to detect at least one portion of the time series in which a measured value is available for the measurement parameter:
assigning flags to the values in the time series of values based on a result of the parsing, wherein a flag assigned to a value obtained at a given timestamp indicates whether the measurement was or was not available at the given timestamp in the time series of values; and
determining whether the special numerical value is a value out of a normal range of values in which the measured values fall or in the range of values,
wherein the determining is based on a comparison between a first count of same-values sequences with the special numerical value in time series of values obtained for the one or more network entities and a second count of same-values sequences with the special numerical value in time series of values obtained for one or more network entities that are shorter than a first threshold,
wherein the special numerical value is determined to be the value out of the normal range of values in response to a ratio between the second count and the first count being below a second threshold, and wherein a flag assigned to a value is equal to a first flag value for each value in the time series of values that is equal to the special numerical value and a second flag value otherwise,
wherein the special numerical value is determined to be the value in the normal range of values in response to the ratio between the second count and the first count being above the second threshold,
wherein the method comprises using a statistical distribution of the lengths of same-values sequences of the special numerical value to detect whether the length of a given same-values sequence with the special numerical value is an outlier in the statistical distribution, and
wherein a flag corresponding to a given timestamp takes the first flag value for each value in the time series of values that is equal to the special numerical value when the length of the same values sequence including the special numerical value is the outlier in the statistical distribution and the second flag value otherwise.
2. The method according to claim 1 , wherein one or more time series of values are obtained respectively for the one or more network entities, and wherein the parsing includes determining if a ratio of a number of values in the one or more time series that are equal to the sole value identified as the special numerical value over a number of measured values in the one or more time series is below a third threshold.
3. The method of claim 1 , wherein a portion of the time series in which the measured value is available for the measurement parameter is detected if the count of value changes occurring in the sliding time window is above a third threshold for at least one temporal position of the sliding time window.
4. The method of claim 1 , further comprising:
performing data analytics on the time series of values based on the assigned flags to generate data analytics results.
5. The method of claim 4 , further comprising:
performing an operation on one or more network devices or a network function based on the data analytics results.
6. The method of claim 1 , wherein the statistical distribution is determined for the lengths of same-values sequences of the special numerical value that are shorter than the minimum length.
7. The method of claim 1 , wherein analyzing the statistical distribution is performed using a classification algorithm to detect a presence or absence of at least one length that is the outlier in the statistical distribution.
8. An apparatus, comprising:
at least one processor; and
at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to,
obtain, for each of one or more network entities, a time series of values of a measurement parameter for respective timestamps, the time series of values including measured values and a special numerical value at one or more timestamps, the time series of values including the special numerical value at a given timestamp for replacing a value of the measurement parameter when no measured value is available for the measurement parameter at the given timestamp;
parse the time series of values to determine which numerical value in the time series of values corresponds to the special numerical value, the parsing including
detecting same-values sequences having a minimum length in the time series of values,
generating a set of at least one value including the value of each of the detected same-values sequences having the minimum length, and if the set of at least one value include a sole value, identifying the sole value as the special numerical value, and
computing a count of value changes occurring in a sliding time window of a given length applied to the time series of values to detect at least one portion of the time series in which a measured value is available for the measurement parameter:
assign flags to the values in the time series of values based on a result of the parsing, wherein a flag assigned to a value obtained at a given timestamp indicates whether the measurement was or was not available at the given timestamp in the time series of values; and
determining whether the special numerical value is a value out of a normal range of values in which the measured values fall or in the range of values,
wherein the determining is based on a comparison between a first count of same-values sequences with the special numerical value in time series of values obtained for the one or more network entities and a second count of same-values sequences with the special numerical value in time series of values obtained for one or more network entities that are shorter than a first threshold,
wherein the special numerical value is determined to be the value out of the normal range of values in response to a ratio between the second count and the first count being below a second threshold, and wherein a flag assigned to a value is equal to a first flag value for each value in the time series of values that is equal to the special numerical value and a second flag value otherwise,
wherein the special numerical value is determined to be the value in the normal range of values in response to the ratio between the second count and the first count being above the second threshold,
wherein the method comprises using a statistical distribution of the lengths of same-values sequences of the special numerical value to detect whether the length of a given same-values sequence with the special numerical value is an outlier in the statistical distribution, and
wherein a flag corresponding to a given timestamp takes the first flag value for each value in the time series of values that is equal to the special numerical value when the length of the same values sequence including the special numerical value is the outlier in the statistical distribution and the second flag value otherwise.
9. A non-transitory computer-readable medium comprising program instructions stored thereon for causing an apparatus to perform a method comprising:
obtaining, for each of one or more network entities, a time series of values of a measurement parameter for respective timestamps, the time series of values including measured values and a special numerical value at one or more timestamps, the time series of values including the special numerical value at a given timestamp for replacing a value of the measurement parameter when no measured value is available for the measurement parameter at the given timestamp;
parsing the time series of values to determine which numerical value in the time series of values corresponds to the special numerical value, the parsing including
detecting same-values sequences having a minimum length in the time series of values,
generating a set of at least one value including the value of each of the detected same-values sequences having the minimum length, and if the set of at least one value includes a sole value, identifying the sole value as the special numerical value, and
computing a count of value changes occurring in a sliding time window of a given length applied to the time series of values to detect at least one portion of the time series in which a measured value is available for the measurement parameter:
assigning flags to the values in the time series of values based on a result of the parsing, wherein a flag assigned to a value obtained at a given timestamp indicates whether the measurement was or was not available at the given timestamp in the time series of values; and
determining whether the special numerical value is a value out of a normal range of values in which the measured values fall or in the range of values,
wherein the determining is based on a comparison between a first count of same-values sequences with the special numerical value in time series of values obtained for the one or more network entities and a second count of same-values sequences with the special numerical value in time series of values obtained for one or more network entities that are shorter than a first threshold,
wherein the special numerical value is determined to be the value out of the normal range of values in response to a ratio between the second count and the first count being below a second threshold, and wherein a flag assigned to a value is equal to a first flag value for each value in the time series of values that is equal to the special numerical value and a second flag value otherwise,
wherein the special numerical value is determined to be the value in the normal range of values in response to the ratio between the second count and the first count being above the second threshold,
wherein the method comprises using a statistical distribution of the lengths of same-values sequences of the special numerical value to detect whether the length of a given same-values sequence with the special numerical value is an outlier in the statistical distribution, and
wherein a flag corresponding to a given timestamp takes the first flag value for each value in the time series of values that is equal to the special numerical value when the length of the same values sequence including the special numerical value is the outlier in the statistical distribution and the second flag value otherwise.