READING TIME 8″
Data is not necessarily information. It could be noise. Do I have to spend millions integrating hardware just to allow for “possible” savings?
Data collection is the onerous organizational effort necessary to collect quantitative and qualitative information on specific variables with the aim of evaluating the parameters and making decisions justified by objective and measurable facts. Good data collection requires a clear process to ensure that the data collected is clean, consistent and reliable.
Determining this process, however, can be complicated. It involves defining your goals, identifying the state of the art in your company, prioritizing actions according to classic budget constraints, deciding on a collection method (and analysis) and finally drawing up an implementation plan for your program. Data can be a very expensive enabler with no guarantees of return.
Define your goals
A crucial mistake we see organizations making is initiating a data collection program without a clearly defined set of forward-looking goals. Data is a means, business is the guiding star: without well-articulated objectives, perhaps SMART, it is very easy to lose course.
Once the key objectives of the project have been established, the next step is to outline how to achieve them. What results do you need to demonstrate to call your project a success?
Identify the requirements for the data
So how can you filter out the noise and focus on what you need? A clear view of your goals will help you narrow your focus and identify a data collection method that exactly meets your needs.
Very often the data that companies really need are few and yet we focus on the massive and indiscriminate collection of the same, with very high costs and machine downtime and the subsequent difficulty of analysis and interpretation.
A paradigm shift is needed, the gold rush of data collection can be less onerous and expensive than you think today.
Let’s imagine applying Occam’s Razor principle. Occam’s (or Ockham’s) razor is a mental model attributed to the Franciscan friar William of Ockham (1287 – 1349), considered one of the most influential philosophers of the 14th century. The principle is defined by William in these terms: “Pluralitas non est ponenda sine necessitate” (Do not consider plurality if it is not necessary) and “Frustra fit per plura quod potest fieri per pauciora” (It is useless to do with more than one can be done with less). The friar wanted to express a criticism of his theory of knowledge of the time which saw a proliferation of new philosophical approaches and explanations. Occam’s razor is a model of thought according to which, all other conditions being equal, the simplest explanation of a phenomenon or the most immediate solution to a problem is always to be preferred: so we must approach data collection.
How to determine your data collection method
The whole purpose of a data collection program is data, but it doesn’t collect itself. There are many options to choose from for how to deal with this problem and each has strengths in a different area. The type of data you are trying to collect, as well as the characteristics of its source and environment, should all indicate which method of data collection makes the most sense for your program.
In the manufacturing field, the first method of data collection is the direct connection of the machines with analysis and measurement systems, read for example PLC, acronym for Programmable Logic Controller, i.e. industrial computers used to control various electromechanical processes to be used in production, in plants. or in other automation environments. PLCs vary in size and form factors. Some are small enough to fit in your pocket, while others are large enough to require you to mount your racks on heavy loads. Some PLCs can be customized with backplanes and functional modules to suit different types of industrial applications.
Making these integrations and measurements has a huge impact on production lines: in terms of costs, in terms of timing (lines must be blocked), in terms of difficulty (many machines, not so modern, cannot be networked at all) .
The second method of data collection uses the IoT. In this case the machines are not connected directly and physically to the detectors, but sensors are also used outside the production lines, which, however, return data that is not contextualized. The lines are not stopped but the problem of analyzing the same data arises.
Tertium non datur, to stay on the subject of medieval philosophy? In this case yes, there is a third way and it is called Screevo. We talk about it below.
The white noise of data collection
Data does not necessarily mean information. Noise is an inevitable problem, affecting the data collection and preparation processes in data mining applications, where errors commonly occur. Noise has two main sources: implicit errors introduced by measurement / collection tools; and random errors introduced by batch or expert processes during data collection, such as in a document digitization process.
The main criticality of data collection through PLC and direct networking of machines is the enormous amount of information accumulated. Machine downtime and the incredible associated cost, combined with the collection of a lot of non-functional data, makes this mode unproductive. It is very likely that a Plant manager and his team will really only need a dozen crucial data. How to balance this with the millions of values amassed?
Not even the IoT is a sensible answer to the needs of manufacturing companies. In this case, despite being “plug and play”, the collection does not return data that are easy to interpret and use.
Screevo, the Voice Process Automation platform that allows the development of voice assistants capable of guiding users through work processes and carrying out data entry activities on their behalf, is the solution to the problem of “white noise” in data collection and it is an economical, scalar, above all “user-centric” method. Let’s see how.
Eliminate the noise
To eliminate the noise it is possible to spend considerable sums or to completely rethink the Data Entry procedures.
Such as?
Through Screevo. Thanks to pure voice interaction, it is possible to ask workers to detect critical data on the line. Do you want to know the temperature of a sensor? The reason for a line stoppage? A reason for rejection? Just ask the line operators.
With Screevo, more data costs just one more question than our voice assistant, and workers can focus on what matters most: creating value with their own hands.
Screevo, through the use of natural language understanding (NLP) and robotic process automation (RPA) algorithms, allows the voice assistant to interact without any difficulty with any software system, eliminating complexity and friction.
What are the benefits?
- Efficient, timely data collection focused on real company needs
- Increased business efficiency, thanks to the reduction of data entry time into the systems
- More sharing: People who speak share more information than people who write
- Data Entry in real time
- Process Mining to support the continuous analysis of possible process inefficiencies
Thanks to Screevo, it will be possible to make the most of the talents and skills of the people present within a production system, allowing them to focus on value-added activities. Screevo revolutionizes the manufacturing sector (and not only) thanks to an approach that frees workers and employees from the slavery of a keyboard or a screen.