Data Quality

The quality of the information is based on the quality of the basic data with which measurements and indicators are built. As data quality we means complete, with valid values (not null), correct type (eg Date, Number), and is in the proper range (zero to N , a valid date, a valid name).

You can help ensure data quality as follows:

By entering certain mandatory fields

Monitoring a "data quality index" using the defined indicator. This indicator is calculated for each data import and individual field level obtained as a ratio between the number of field values considered correct vs. the number of values considered incomplete

Fig. 22. Example of variation of quality indicator data entry

Process Control

You can monitor the process performance using indicators and measurements, according to their business objectives. These indicators are constructed from basic measurements with high reusability in different contexts and processes: size, effort, reuse, productivity, errors, rework. Have access to a rich set of indicators covering a wide range of processes (management, engineering and process support). Through the use of "process control graphics" can analyze data generated at each iteration of project over time. These graphs allow discrimination "noise" from "signal" using very simple statistical techniques (based on a "normal" distribution, so the data must be "homogeneous"). Given the amount of data available it is important to discriminate information from simple "noise". This can be viewed directly "statistical signal" in the data to determine points in a data series that require "explanation" of its assignable cause variation. These cases are characterized by an unexpected change in the performance of the process. Also known as "assignable causes" because they can be identified, analyzed, and taken into account to prevent recurrence. May be due to:

Pressure dates
Lack of adherence to process
Planning of tasks on the thread
Uncontrolled inputs (eg, load error)
Errors "sampling"

You can employ two types of curves to display graphics signals on the variation of the data, as well as the magnitude of each variation:

Graph X: graph the variation around the average value of indicator values (+-3sigma bands indicating change)

Graph mR: the magnitude of change for each point

To improve the performance of your processes, the categorization of the variation in the data reported by the indicators is the first step to take. You can discern between different variation patterns, each with its implication including:

Points "dominant": a possible signal process (or fault data) that cause an overflow outside the natural statistical limits, given by + - 3 sigma variation from the mean

Points "moderate": signal a possible change process that moves the average natural boundary near the bottom or top

Points "weak": signal a possible process that causes a shift down, or up, the average

Points "near average": a possible signal that keeps the process close to the average performance

Points "ascendant": signal a possible change process that moves something the average downward or upward

As reality is complex and not everything in it is "white" or "black", you need a tool that provides a vision "nuanced" close to common sense. This can display signals such "fuzzy" in the data, complementary statistics. The technique differs from traditional logic in which an item can belong more than one set, ie its degree of membership is a value ranging from 0 (does not belong) to 1 (belongs "totally "), unlike traditional logic, where the membership value is 0 or 1. Thus, this logic can be used qualitatively close to nature languages, common sense, as "little", "high", "much", etc.. You can employ qualitative signals for:

Rate the level of statistical signal information (entropy level, or "temperature" of the data), to discriminate unstable stable processes incorporating the nuances

Rate any indicator in relation to a consolidated group organizational, to compare variations in qualitative terms, such as a sign identifying "diffuse" the change of ownership of a set of indicator value (considering their degree), for example the transition from set "very low" to qualify as "quite high"

Rate projects by their degree of belonging to different sets of volume of activity, to filter the repository for those projects that have more or less regardless of type activity

Measurements analyst has all the information necessary to determine past performance, current and future estimated that each indicator provides so we can support better decision-making and continuous improvement of processes available through graphics, rules of pattern analysis variation, filters and other facilities like comparing baseline and extrapolation or comparing individual series. You can access this type of analysis directly from the tools

Fig. 23. Distribución de datos normal

The Context

The context is the quantitative / qualitative information associated with a graph / data of an indicator for a project / group in a period of time that gives it meaning and that it must always accompany them in order to be feasible its interpretation. This context is generated automatically by the application.

A summary of the numeric context information can be viewed directly on the data table. For example:

"Goal: Measure the number of defects in the process. Problems to know the quality level of the products elaborated and make the necessary adjustments to the process Formula: Ratio of defects found in Resol. Problems vs. total defects found QualityIndex: 0.6606 Created : 27 April 2017 RepositoryDate: 27 October 2016 12:03:04 am "

The complete context can be visualized with an inspector, and in addition to the numerical information it contains statistical graphs that allow analyzing the main graph of the panel, such as frequency histogram (sigmas), class intervals, re-scaled / Pareto range according to the type of graphic

Among the numerical information is available:

Average value of the data
Standard deviation
Quantitative / qualitative dispersion (gini)
Quantitative / qualitative diversity (theil)
Normal test
Temperature: qualitative according to the amount of information (signals) present in the data
Name of the procedure for obtaining the data
Indicator Formula
Quality Index
Repository Date
Project Type
Objective of the indicator
Algorithm used to obtain the data (SQL / IHDSL)
Project performance (qualitative)
Indicator process
Indicator unit of measure
Date of creation of the indicator
Indicator description

The components of the inspector context can be copied to the paste buffer so that they can be pasted into a sheet of objects on the board or on the desktop as individual elements (although they should always accompany the original data / graph in order not to lose meaning)

There are several types of context. A type of context, called "Global" is a set of events (dates) on a timeline. A special event is "Source" and can be used to filter data from that date. This context is generated and maintained by the user through "Global Notes".

The "Annotations" allow to create specific context, associated to points of a line graph, for example to indicate possible explanations of causes of statistical variation of the data directly on the graph. These annotations can be viewed / hidden as labels. Other types of automatically generated labels indicate context of statistical and diffuse variation (signals)

The context is maintained by adding an indicator to the Board, and is persisted in the baseline of the same

Fig. 24. Context associated with a particular indicator

Activity Systems

In Improvekit we return to the primordial philosophy of object orientation, which is to be based on mental models. In the framework of the system, we do it using the Activity Systems. Activity theory helps explain how social artifacts and social organization mediate social action.

The goal of Activity Theory is understanding the mental capabilities of a single individual. However, it rejects the isolated individuals as insufficient unit of analysis, analyzing the cultural and technical aspects of human actions. Activity theory is most often used to describe actions in a socio-technical system through six related elements of a conceptual system expanded by more nuanced theories:

Object-orientedness : the objective of the activity system. Object refers to the objectiveness of the reality; items are considered objective according to natural sciences but also have social and cultural properties.

Subject or internalization : actors engaged in the activities; the traditional notion of mental processes

Community or externalization : social context; all actors involved in the activity system

Tools or tool mediation : the artifacts (or concepts) used by actors in the system. Tools influence actor-structure interactions, they change with accumulating experience. In addition to physical shape, the knowledge also evolves. Tools are influenced by culture, and their use is a way for the accumulation and transmission of social knowledge. Tools influence both the agents and the structure. In improvekit we also consider tools as "edge objects" (entities that can link communities together as they allow different groups to collaborate in a common task)

Division of labor : social strata, hierarchical structure of activity, the division of activities among actors in the system

Rules : conventions, guidelines and rules regulating activities in the system

The data sources, and certain indicators, contain metadata to represent the components of the related activity system, in particular, the instruments, operations and intervening actors.

Activity patterns

Self-constructed graphical formalism which is a socio-technical systemic view that relates the following components in a solution pattern from the signals and other clues in the data:

context: the general context present in the story, derived from each indicator

        Summary of the general context:
            Tasks with duration exceeded: measure the volume of tasks that exceed the established time limit to determine bottlenecks and corrective actions.
            Task Type Activity Volume (Monthly) - Measure task size regardless of effort (quarterly).
            Activity Volume: Measure project performance based on estimated task size.

the problem (as contradictions / tensions): problems can be expressed as contradictions / tensions in the activity system. The contradictions are entered by the user (a series of possible defaults contradictions derived from the data are suggested, including internal stresses to a component of the system, stresses between two components, due to changes, with other activity systems, or predefined from parameters (such as baseline thresholds, number of actors, roles, workflow, etc.)

        AS Tasks with exceeded duration @ Improvement of Processes and Tools
        AS Task type Activity Volume (monthly) @Improvement of Processes and Tools
        AS @ Activity Volume Process and Tool Improvement

the causes: they are defined based on the patterns detected in the data, such as statistical signals, diffuse signals, etc.

        Tasks with duration exceeded: measure the volume of tasks that exceed the established time limit to determine bottlenecks and corrective actions.
    Process stability
        Stable process Values â€‹â€‹fluctuate around an average of 34.0000 with September 13, 2019 10.0000 and maximum minimum values â€‹â€‹January 17, 2020 68.0000. it is not normal distribution. The kurtosis is of the peak type and straight symmetry.
        Patterns
            The values â€‹â€‹have a "slightly low" degree of dispersion and "0..2" diversity.

questions, arguments: defined by the user, for example from situational questions (what, how, when, where)

possible solutions: defined by the user, for example from past experiences of innovation

It is used in Story to explain / argue the set and is based on the theory of Activity Systems and Patterns

Fig. 25. An activity system under analysis

Concepts

Data Quality

Process Control

The Context

Activity Systems

Activity patterns

Categories