This document discusses improving inventory management through effective data modeling and metrics. It recommends identifying the most valuable metrics, collecting day-to-day data to provide managers insights, and reducing loss. Effective dashboards and key performance indicators using multi-dimensional data analysis can help improve business decisions and ensure profitability. Proper data collection, storage, exploration and reporting are also important for inventory management.
2. • Inventories can easily
represent 40-50% of a
company’s capital
investments.
• Without the right tools
& strategy inventory
management might
turn into a nightmare.
3. Identify the most
valuable metrics
Day to day data
Give managers the
insights they require.
Reduce loss.
11. • A multi-dimensional
dataset.
• analyzing data to look
for insights.
• Each cell of the cube
holds a number that
represents some
measure of the
business, such as sales,
profits, expenses,
budget and forecast.
12. Too many KPIs and often
the wrong
ones
How effective are your dashbords ?
13.
14.
15.
16.
17.
18. Effective dashboards and KPIs can help improve
business decisions, hasten corrective actions and –
hopefully – ensure balanced and sustained
profitability
Material management could have a very sharp cutting edge. As you can see, inventories can represent up to 50% of a company’s capital investments, which is a pretty huge number.
So obviously, without the right tools and strategy, material management might turn into a nightmare to the company as well as the managers.
So during this whole semester, we worked on identifying the most valuable metrics, gathering day to day data, so that we can give the managers the insights they require in order to reduces losses, and by losses we mean, data losses as well as material losses.
To do it well, we should start with modeling our project according to the needs and this must be done with focusing on relevant data among others to make sure to get the beneficial needed results,
Let’s go deeper with explaining some technical terms
Fact: is the structure where we mention the measures that we’re going to use
Dimension: is the structure where we identify analysis axes referring to the measures
One the model is ready, we move to implement it with ETL which is the abreviation of Extract Transform Load
In our project we did in first place the Operational Data Store (ODS) where we only extract and load all type of data from all type of sources, we used the ODS package to ease the system,
After that we implement the DW based on the model we’ve already prepared,
So it’s cool to exclude the excel files and collect all the Data from them , centralize it and clean in them grouping them to get informations in DW and Analyse them in the OLAP Cube, but you know what is cooler, it’s the magic of reporting, Wann know why ? Too many KPIs doesn’t mean that they are the right way ti get a good picture about our company, and know we will show you how reports makes the difference,
In this map, we have a visibility about all the countries that deals the most with BigConsult judging on the quantity of product imported from them. From the map we can clearly notice that The USA and Germany are the most countries dealing with Big Consult.
In this figure, we have a table grouping all the countries dealing with BigConsult, whenever the responsible will click on a country, he will be redirected to another interface where he’ll get a chart presenting the quantity of products (Kg) per Land.
Within this figure, we have information about all the deliveries of the companies dealing with BigConsult, we tried to calculate the delay between the order date and the delivered date in order to help the responsible to qualify a provider if he is respecting the delays « green indicator », or if he is little bit late « orange indicator » or if his high delays can occurs some losses for the company « red indicator ».
With this figure, we can notice the amount of money spent to buy each product from a specific supplier and evaluate it according to its importance then evaluate the supplier if he is a full star one (important provider) or less than that.
In this figure we have a visibility of all the suppliers of a specific product and besides it, there is a chart showing the top 3 providers with its percentage.
Every day we generate quintillions of bytes of data. An important amount of this data is user-generated social media content.
Since we live in a digital age where everybody express their thoughts, feelings & behaviours in social media
Businesses figured out that measuring their social media outreach will help them unlock insights from their customers generated content.
However, data gathered from social media like facebook & twitter is not typical data that we can store in relational database.
First, we collected data using web scrapping methods, we stored it in NoSql database like Mongo. In the 3rd step we used R to explore what we have
And connected our datasource to Tableau software to generate the following reports
First of all, we wanted to get an idea about users who interact the most with SAP social media profile, where they live and which language they speak in order to publish more targeted posts.
Tweets with the most impact usually get retweets wich increase their visibility and engage more twitter users, but sometimes users opt to like tweets which is an other popularity indicator
Visualizing the facebook page activities on a monthly basis can help page administration identify the most appealing posts and how the interact with them.
In order to have a classification of suppliers from a given criteria, we applyed the Agglomerative hierarchical clustering technique on our data warehouse.
This correlation circle, which is the result of the Principle Component Analysis, gives us a little idea about how our axes will present the information depending on the correlation of its variables.
As you can see, there is a similarity between some group of suppliers depending on their country. We could make a group of English speakers (High price), EG: Arabic (low price), and the rest of the world (average price).
After using the classification method (CAH), we got this dendrogram. We noticed that we can get 3 classes by cutting it horizontally. Grouping US (4 provider), EG (1provider) and the rest of the word which approves the previous results with the (ACP).
In this part, we tried to gather information from the social media in order to get some feedbacks and have an idea about what our users are saying about us. In the first place, we tried to pick up the most used words or terms in the tweets related to us. Then we converted the terms into topics , and as you can see in this chart, this is the topic’s use frequency according to the time dimension. Next we made a sentiment analysis by studying the polarity of the tweets in order to know wether the tweets are positive or negative. That being said, we’ll manage to get some feedbacks about the degree of satisfaction of our customers.