Temporal Data and Weather Forecast

Weather data is temporal data. It changes according to time (hours, day, month or years). Since weather is crucial for us human to do our activities, people forecast it. By having forecast data, it helps business and people plan for their outdoor activities. In Malaysia – it is very crucial during kenduri kahwin, some people still believe and hire ‘bomoh hujan’ to forecast and prevent rain during the wedding day. More over, it helps business plan like transportation, construction and farmers for crop irrigation and protection. Eventhough the weather data in Malaysia is not as crucial as in four season country (since it will not help people on how to dress or either to bring extra coat for windy days) but forecast data can help in term of health issue like asthma and heat stress especially for children’sschool activities.

Since the data for Forecast weather is everywhere – from your own handphone, PC, TV and radio. I think for Haida (since you are from MANA – assurance course), it is time for us to check the accuracy between the forecast data and the real one. It will help to prove the accuracy of Jabatan Meteorologi Data. If the comparison has been done, why dont we visualize the comparison to ease the forecast data understanding.

Thus, the objectives for Haida research can be something like this

  1. Compare between forecast and real set of weather data.
  2. Visualize the comparison
  3. Identify the accuracy of the forecast data – reliability/assurance of the data OR maybe we can access how people trust/ the data?  (hmm.. this can give awareness about the credibility of jabatan meteorologi data)

In order to do that, what you need to do this week is:

  1. Identify and get the forecast and real set of weather data (try to get a set for 10 days first)
    1. Type of data – the general one, things like this:

Source: weather underground weather forecast.

  1. Bring that set of data for our next discussion (17 August 2017).
  2. Have a peek on your expected outcomes, something like this (but not necessarily exact):

  1. Read and understand this http://stat4701.github.io/edav/2015/03/25/hafiz-weather-1/very good to get some ideas for your LR (please explain to me your understanding about this article in our discussion later)

How is telecom industry benefiting their data?

How is telecom industry benefiting the big data?

Telecom industries are sitting on a gold mine, as they have plenty of data. But what they require is a proper digging and analysis of both structured and unstructured data to become a valuable asset to the industries.

Big Data from the perspectives of telecommunication industry

Through proper digging, they are able to get deeper insights into customers’:

  • Behaviour – combat fraud
  • Service usage patterns – marketing interest, marketing agility  (related to temporal data)
  • Preferences
  • Real time interests – real time customer insights (related to temporal data)

From Acker et al (2013), the telecom industry must experiment their own data. Demonstrating what they have on hand to see what kinds of connections and correlations it reveals, This process must be carried out iteratively to emerge the more efficient operations and more effective marking.

Source: Acker et al (2013)


  1. http://bigdata-madesimple.com/11-interesting-big-data-case-studies-in-telecom/
  2. Ackers (2013) Benefiting from big data. A new approach for the telecom industry. published by Booz & Company.

Temporal Data and Business Intelligence

Temporal Data and Business Intelligence

According to Aigner et al (2007), time is an outstanding dimension. In popular physics, time is the fourth dimension. For ages, scientists have been thinking about meaning and implications of time. Understanding temporal relations enables us to learn from the past to predict, plan, and build the future. In the world of computerization, time is like an invisible presence. Business systems mostly operate in some sort of existential present tense and programmed into a unique timestamp, flags and update. Hence, it is no surprise that time is also a key concern in Visual Analytics, where the goal is to support the knowledge crystallization process with appropriate analytical and visual methods [1]. Visualizing time-oriented data, which is the focus of this paper, is not an easy business. Even though many approaches to this task have been published in recent years, most of them are specific to only a particular analysis problem. The reason why most methods are highly customized is simple: it is enormously difficult to consider all aspects involved when visualizing time-oriented data.

In current years, the temporal data is still valid and significant to emphasize the value of Business Intelligence (BI). The temporal concept enables business users to explore past business events to understand problems, see trends (monitor and review the data) and these activities can support the decision making as well. Thus, in many cases, users get point-in-time views of the business at defined times, such as the daily or month-end close. However, as business operations have become increasingly real-time in nature, the need for a continuous history that provides an ongoing, complete and accurate record of transactions and business performance has become paramount. Temporal — and in particular, bi-temporal — data is thus central to effective BI processes and should be a core part of any data warehouse or data mart (Devlin, BI Expert Panel)

Furthermore, the business context is growing in the complexity of both BI needs and the temporal characteristics of data. Decision making is increasingly real-time in nature. Predictive analytics extends our interest from the past and present into the future. Big data and the Internet of Things take us into a new world where the people who use data often don’t control its structure or content. Even the bi-temporal model, Johnston is reluctantly forced to admit, may turn out to be insufficient to carry all the temporal meaning that a user may wish to impart to or extract from business data. A tri-temporal model, with three-time axes, may eventually be required to make full sense of data. Then quoted from Johnson, the understanding for the interconnection between Bi, analytics and data temporal is vital if we are to navigate the expanding world of big data. Without such an understanding – Devlin believes the implementation of BI was destined to crash and burn. Therefore, it should be seen as mandatory preparation to revisit the theory and practice of temporal data.

An important issue that concerns all previous points is task-orientation. This means that Visual Analytics systems should automatically suggest and parameterize visual, analytical, and interaction methods based on the users’ task at hand. Recently, an interesting analysis of possible visualization tasks has been published in [3]. That list of tasks can be used as a basis for future research on task-oriented Visual Analytics. In that regard, perceptual issues must be further investigated. Empirical tests have to be conducted to judge which forms of presentation (2D or 3D, static or dynamic, etc.) are best suited for particular analysis tasks.

Aigner, W., Miksch, S., Müller, W., Schumann, H. and Tominski, C., 2007. Visualizing time-oriented data—a systematic view. Computers & Graphics31(3), pp.401-409.

Devlin, B. Temporal Data Reality: In BI, time is of the essence. https://goo.gl/VdEt7T

Representation of Data Monitoring – Part 2: How

  1. Timeline:
    Line drawn on a suitable scale (days, month, years, centuries) on which key historical, planned or projected events and perionds marked in the sequence of their occurence (an incident or event; the fract or frequency of something happening, the fact of something existing or being found in a place or under a particular set of conditions) – this bring me to temporal data.