ENG
engineering & construction
The role of Big Data in the evolution of the engineering and construction market
Artificial Intelligence, IOT, Machine Learning, BIM and many other big technological trends have revolutionized the construction market, and they all have a common background: Big Data. This is because they need or feed on data to generate insights, prediction, security, transparency and efficiency to engineering and construction processes.
Therefore, creating well-established processes to collect, store and use data in a strategic way is the first step to take advantage of all the benefits that these innovations can offer. In addition, including data collection, processing and analysis in construction management is crucial to prepare your company for the future and ensure that it remains competitive in the market.
Thus, there is no point, for example, in investing in BIM platforms and artificial intelligence tools if your company does not have the necessary information to make these technologies reach their full potential.
How do you know if patients will miss their appointments?
It's very bad to wait for clients who don't come, isn't it? Besides bringing financial losses, it leaves your team idle, since, by separating a certain time for a client, you lose the possibility of serving other patients in the same period. These absences disorganize the routine, compromise the planned budget and result in idle points in the agenda that end up generating financial losses at the end of the month.
Did you know that there is a way to predict who is most likely to miss your appointments and thus not lose money?
Missing appointments is a recurring problem that can greatly hinder the performance of your entire team. After all, it is uncomfortable to prepare to receive a patient, reserve part of your time for this and, in the end, he doesn't show up. Of course, we all know that we all have our commitments and, many times, the unexpected happens. However, the unforeseen is not the only reason why appointments are often missed. There are a number of external (independent) factors that influence client absences and many of these can be used to somehow find a way to predict the chance of a client missing a scheduled appointment and thereby control and reduce absences and delays to pre-scheduled appointments.
How does it work?
To do this, it is necessary to analyse the client base and appointments made and identify patterns of absences relating to social, socio-demographic, age, procedure performed profile, as well as external factors such as those relating to the weather. This data is then georeferenced, enriched and statistical models and simulations are applied to it in order to calculate the probability of absences in a given period.
Are you interested and want to know more about how to use artificial intelligence and big data to predict who will miss your appointments and thus avoid future holes in your schedule and reduce costs?
Talk to us! We have developed an algorithm to predict the probable number of absences from a set of individuals with 80% accuracy.
Data Lake: Solving the data access problem in large companies
What is a data lake anyway?
A data lake is a repository that allows the storage of a large amount of raw data in its native format, including structured, semi-structured and unstructured data, which brings several benefits to those who adopt it, such as
Centralisation: data from multiple sources is centralised in a shared location;
Scalability: it allows the storage of a large amount of data and has the potential to expand its size as the amount of data increases;
Low storage cost: storage costs are a major concern that needs to be taken into account. The data lake offers low cost storage for the data;
Variety of data: data lake has the ability to store different types of data like transactional data, API data, sensor data, binary data, social media data, chat, etc;
Zero bureaucracy: users from various departments can quickly access data lake content as it is stored in a central repository. As a result, a user can easily collect data considered important to drive business decisions in any area.
When the source data is in a data lake, without a single structure or control schema built in - in a data lake the data structure and requirements are not defined until the data is needed - new case support can be much more direct and quicker, allowing employees to access any information needed and thereby develop the reports they want, using the tools they want. In this way, IT becomes the custodian of the infrastructure and data in the cloud, while others take responsibility for exploring and extracting it.
Since the value of the data is not clear from the start, it is not classified at the time it is stored in the data lake, it is loaded in its raw format and made available for use. Only when the data is accessed is it then classified. As a result, costly data preparation is eliminated. Furthermore, since the cost of data storage is minimal and large volumes of data can be stored at any time, there is no need to decide which data is relevant, you can just store all the data in the data lake. Not least because data that seems insignificant at the moment may become significant in the future.
In short, a data lake acts as an intelligence centre for companies, a unified database whose purpose is to be ready for an unknown usage need.
However, it is important to note that a centralised data warehouse is only useful when stored data needs to be extracted for use by different departments. In addition, to deploy a data lake at the enterprise level, certain features are required to enable its integration into the overall IT application and data management strategy, as well as into the organisation's data flow landscape. Also, it is extremely important to ensure that the data lake is getting the right data at the right time. Therefore, integration platforms that operate in the background must be able to send data from various tools, in real time and on demand, based on different business cases.
Moreover, a data lake is not just about storing data centrally and delivering it according to different departments. With more and more users starting to use it directly or via applications and analytical tools, the importance of governance for the data lake increases. In this context, the main challenge is to ensure that data governance policies and procedures exist and are enforced. There should be a clear definition of the owner for each dataset as well as information on how and when this data enters the data lake. Everything needs to be very well documented regarding the accessibility, integrity, consistency and updates of each piece of data, involving tracking and logging of the manipulation of data assets present in the data lake, based on well-defined policies and guidelines.
In large enterprises, perhaps the most powerful impact of a data lake is the activation of innovation. Since the technology has the potential to help break down information silos and other barriers. Giving managers a clearer picture of the business enables them to understand the constraints between functional units and facilitates collaboration, which can, in the long run, transform the culture of the business.
more efficient stocks with A.I.
If you buy more items than you need, you will end up with items lying unused in your stock, i.e. you tie up capital unnecessarily, which leads to storage and maintenance costs. If you don't buy enough items, you will quickly run out of stock. And an unavailable product that is not in demand is not only a lost sale, but also an unhappy customer who may choose a competitor the next moment. For this reason, it is extremely important to understand the behaviour of your sales and what your future demand will be.
Forecasting demand and inventory
Forecasting methods use analysis of future projections of past data and other factors that affect outcomes, such as seasonality, lead times, production errors and market changes. To help determine planning cycles, production and inventories to facilitate decision making. Companies that do not pay much attention to their forecasting or do not do it correctly may face problems such as surplus or shortage of inventory, wastage of raw materials, unused capacity, etc. For this reason, forecasting plays a fundamental role in business strategy.
For example, demand forecasting allows companies to estimate future demand and determine how the operational and production process will take place, transforming raw materials into a final product for consumers. In short, demand forecasting means determining the quantity of raw materials to be purchased, the quantity of products to be manufactured, the number of products to be delivered, the number of employees to be hired and the number of facilities to be built.
Being able to estimate the amount of products a shop will sell in the future, i.e. forecasting sales, will enable shop owners to determine the inventory required to avoid a surplus or shortage of items. Management should therefore pay close attention to this process by matching the sales forecast with the shop's inventory and putting together the right product mix, as the sales forecast directly affects future costs and profits.
Similarly, inventory forecasting models are also critical elements of the forecasting process. For wholesalers and distributors of durable goods, inventory forecasting is particularly important as it forms the basis for all business plans in terms of market and sales forecasts.
Classical forecasting methods vs. machine learning models
However, forecasting is fraught with uncertainty because most events that affect the market simply cannot be modelled deterministically. For example, it is difficult to predict a price cut by competitors or an event that prevents a supplier from delivering its products on time. It is even more difficult to predict the emergence of a superior new technology. And this is where solutions based on machine learning come into play. Classical forecasting methods, such as average, periodic and uniform time series forecasting, almost completely eliminate this uncertainty. Since uncertainty is ignored in whole or in part, many situations may simply not be reflected in the forecasts at all. Nevertheless, this type of forecasting is still prevalent in companies.
To address this uncertainty, it is desirable to predict not only the most likely future outcome, but also other alternative outcomes. Probabilistic forecasting is the most popular statistical formalisation of this insight. It produces a statistical estimate for all possible outcomes, thus covering a much wider range than classical methods.
In addition, most of the techniques currently used are subject to significant forecasting errors. This leads to difficulties in identifying trends in the data and a limited ability to understand the causes of variability. However, the classical techniques are based on time series forecasting, which can only take into account a few future demand, sales and inventory factors. However, forecasting is not just about future values. These are just a few of the many elements that need to be forecast.