Tutorial#

Introduction#

This tutorial shows the steps for creating and instantiating a Dlite entity, and populate it with data. For this, we will use an example csv file containing data corresponding to time measurements from a solar panel. The example file has the following format:

time	Impp [A]	Isc [A]	MPP [W]	Vmpp [V]	Voc [V]
2022-08-01 00:41:00	1.188	1.25	38.182	32.14	37.33
…	…	…	…	…	…

Writing a DLite Entity#

Let us write an Entity representing our solar panel measurement. We start by creating a json file called myEntity.json.

Step 1: Giving it a unique identifier#

Firstly, we must provide a unique identifier for our Entity. There are several ways of doing this:

1. Uniform Resource Identifier (URI)#

We can create and assign a unique Uniform Resource Identifier (URI). A URI has the form namespace/version/nameOfEntity, and it can be either written directly in the json file:

"uri": "http://www.ontotrans.eu/0.1/solarPanelMeasurement",

or we can give the namespace, version, and name of our Entity as separate fields:

"name": "solarPanelMeasurement",
"namespace": "http://www.ontotrans.eu/",
"version": "0.1"

2. Universally Unique Identifier (UUID)#

Alternatively, we can generate and assign a universally unique identifier (UUID). In Python, this can be achieved by the use of the dlite.get_uuid() function or the dlite-getuuid tool.

import uuid

# Generate a random UUID
entity_uuid = uuid.uuid4()

with open("myEntity.json", "w") as entity_file:
    entity_file.write('"uuid": {}'.format(entity_uuid))

Our json file now contains the following:

"uuid": <uuid>

where <uuid> is a unique identifier for our Entity.

Step 2: Providing a link to the metadata#

Let us assume that we have followed option 1 and provided a URI. The next step consists in adding a meta field that will link the Entity to its metadata. In DLite, all Entities are instances of the EntitySchema. The EntitySchema defines the structure of the Entity. In the meta field of our Entity, we need to provide the URI of the EntitySchema. For a schematic overview of the DLite data structures and further detail, see Concepts section of the DLite User Guide. We add this information to our json file in the following manner:

"uri": "http://www.ontotrans.eu/0.1/solarPanelMeasurement",
"meta": "http://onto-ns.com/meta/0.3/EntitySchema",

Note

The meta field actually defaults to http://onto-ns.com/meta/0.3/EntitySchema. This means that the meta field is optional for entities.

Step 3: Adding a human-understandable description#

Next, we want to include a human-understandable description of what the Entity represents. In our example case, such a description field could be

"uri": "http://www.ontotrans.eu/0.1/solarPanelMeasurement",
"meta": "http://onto-ns.com/meta/0.3/EntitySchema",
"description": "Measurement data from one solar panel."

Step 4: Defining the dimensions#

We continue by defining the dimensions for our Entity. Each dimension should have a name and a human-understandable description. In our example case, we only have one dimension since all the quantities we are interested in are measured with the same frequency (i.e., we have the same number of data points for all the measured quantities). Note that even though we only have one dimension, we need to give it in a list format.

We will give our dimension the generic name “N”, and describe it as the number of measurements:

"uri": "http://www.ontotrans.eu/0.1/solarPanelMeasurement",
"meta": "http://onto-ns.com/meta/0.3/EntitySchema",
"description": "Measurement data from one solar panel.",
"dimensions": {
      "N": "Number of measurements."
}

Step 5: Defining the properties#

Now it is time to define the properties of our Entity. Here is where we can give meaning to our data. As for the dimensions, we add the properties as a list of json structures, each one having a name, type, description, and if relevant, unit, shape (the dimensionality of the property) and ref (the metadata for ref types).

Inserting the properties displayed in the table above, our Entity is complete and may look like

{
    "uri": "http://www.ontotrans.eu/0.1/solarPanelMeasurement",
    "meta": "http://onto-ns.com/meta/0.3/EntitySchema",
    "description": "Measurement data from one solar panel.",
    "dimensions": {
        "N": "Number of measurements."
    },
    "properties": {
        "t": {
            "type":"str",
            "unit":"ns",
            "shape": ["N"],
            "description": "Time"
        },
        "MPP": {
            "type":"float64",
            "unit":"W",
            "shape": ["N"],
            "description": "Maximum Power"
        },
        "impp": {
            "type": "float64",
            "unit": "A",
            "shape": ["N"],
            "description": "Maximum power point current."
        },
        "isc": {
            "type": "float64",
            "unit": "A",
            "shape": ["N"],
            "description": "Short circuit current."
        },
        "vmpp": {
            "type": "float64",
            "unit": "V",
            "shape": ["N"],
            "description": "Maximum power point voltage."
        },
        "voc": {
            "type": "float64",
            "unit": "V",
            "shape": ["N"],
            "description": "Open circuit voltage."
        }
    }
}

Note

Both dimensions and properties can also be provided as arrays using a dict with the name as key. This may look like

    "dimensions": [
        {"name": "N", "description": "Number of measurements."}
    ]
    "properties": [
        {
            "name": "t",
            "type":"str",
            "unit":"ns",
            "shape": ["N"],
            "description": "Time"
        },
        {
            "name": "MPP",
            "type":"float64",
            "unit":"W",
            "shape": ["N"],
            "description": "Maximum Power"
        },
    ...
    ]

DLite supports both syntaxes.

Loading an Entity with DLite#

We will now load our Entity in Python. There are several ways of doing this.

1. Using the json storage plugin:#

import dlite

SolarPanelEntity = dlite.Instance.from_location('json', 'myEntity.json')

2. Adding the file to `storage_path`#

In this method, we append the json file to storage_path and then use the get_instance() function to load the Entity using its URI. In Python, this may be done by

import dlite

# Append the entity file path to the search path for storages
dlite.storage_path.append('myEntity.json')

# Load Entity using its uri
SolarPanelEntity = dlite.get_instance("http://www.ontotrans.eu/0.1/solarPanelMeasurement")

Note: The same approach applies if a UUID is used.

Connect loaded entity to data#

Now that we have loaded our Entity, we can instantiate it, and populate with the data from the csv file mentioned in the introduction. Note that you may have to provide a full file path to the read_csv() function, depending on where you have stored your csv file.

import pandas as pd

# Read csv file using pandas, store its contents in a pandas DataFrame
solar_panel_dataframe = pd.read_csv('data.csv', sep=',', header = 0)

# Instantiate an instance of the Entity that we loaded above
inst = SolarPanelEntity({"N": len(solar_panel_dataframe)})

# Create a dicitionary that defines the mapping between property names and column headers
mapping = {
       "t":"time",
     "MPP":"MPP [W]",
     "voc":"Voc [V]",
    "vmpp":"Vmpp [V]",
    "isc" :"Isc [A]",
    "impp":"Impp [A]"

}

# Loop through the dictionary keys and populate the instance with data
for key in mapping:
    inst[key] = solar_panel_dataframe[mapping[key]]

More examples of how to use DLite are available in the examples directory.

Working with physical quantities#

As you can read in the previous sections, a property of a dlite entity can be defined with a unit (a unit of measurement). The dlite.Instance object can set (or get) their properties from (or to) physical quanities (the product of a numerical value and a unit of measurement). The dlite library uses the physical quantities defined and implemented by the Python package pint.

You must install pint if you want to to work with physical quantities, try to run the command pip install pint, see the page pint installation.

This section illustrates how to work with physical quantities when manipulating the dlite instance objects. The code example will use the following data model:

uri: http://onto-ns.com/meta/0.1/BodyMassIndex
description: The body mass index of a person.
properties:
  name:
    type: string
    description: Name of the person.
  age:
    type: float32
    unit: year
    description: Age of the person.
  height:
    type: float
    unit: cm
    description: Height of the person.
  weight:
    type: float
    unit: kg
    description: Weight of the person.
  bmi:
    type: float
    unit: kg/m^2
    description: Body mass index.
  category:
    type: string
    description: Underweight, normal, or overweight.

The code below will create one BodyMassIndex instance object (variable person) and assign the properties of the person using physical quantities. You can use the method person.set_quantity to set a property with a quantity or you can use the shortcut person.q like shown below:

BodyMassIndex = dlite.get_instance("http://onto-ns.com/meta/0.1/BodyMassIndex")
# create a BodyMassIndex instance
person = BodyMassIndex()
# assign the name of the person
person.name = "John Doe"
# assign the age of the person without a physical quantity
person.age = 5
assert person.age == 5.0
# assign the age of the person with a physical quantity
person.set_quantity("age", 3.0, "years")
assert person.age == 3.0
person.q.age = (1461, "days")
assert person.age == 4.0
# assign the height and the weight
person.q.height = "1.72m"
person.q.weight = "63kg"
assert person.height == 172.0
assert person.weight == 63.0
# compute the BMI
person.q.bmi = person.q.weight.to("kg") / person.q.height.to("m") ** 2
assert np.round(person.get_quantity("bmi").magnitude) == 21.0
# the following line will give the same result as the line above
person.q.bmi = person.q.weight / person.q.height ** 2
# the following line will raise a TypeError exception, because the property
# "category" with the type "string" cannot be converted into a quantity.
category = person.q.category
# assign the category
person.category = "normal"
if person.bmi < 18.5:
    person.category = = "underweight"
elif person.bmi >= 25.0:
    person.category = "overweight"

If you need in your program to use a specific units registry, use the function pint.set_application_registry

The Python property q of the dlite.Instance objects as some other methods:

assert person.q.names() == ["age", "height", "weight", "bmi"]
assert person.q.units() == ["year", "cm", "kg", "kg/m^2"]
for name, quantity in persion.q.items():
    quantity.defaut_format = "~"  # format the unit with a compact format
    print(f"{name} = {quantity}")

see quantity formatting

You should not keep the Python property q as a local variable, it will result with wrong assigment if you work with several instances.

# >>> don't do that
p1 = person1.q  
p2 = person2.q
# this will assign the weight to person2, and not to person1
p1.weight = "34kg"  
# <<< 
# you must write this code
person1.q.weight = "34kg"
person2.q.weight = "32kg"
# for more complex formula, you can get the quantities in local variables
w1, h1 = person1.q.get("weight", "height")
w2, h2, a2 = person2.q.get("weight", "height", "age")

The first line of the code above prepare the singleton dlite.quantity.QuantityHelper for the person1 and return the singleton. The second line do the same as the first line, so the variable p2 is the singleton and is prepared to work on person2.

Tutorial

Contents

Tutorial#

Introduction#

Writing a DLite Entity#

Step 1: Giving it a unique identifier#

1. Uniform Resource Identifier (URI)#

2. Universally Unique Identifier (UUID)#

Step 2: Providing a link to the metadata#

Step 3: Adding a human-understandable description#

Step 4: Defining the dimensions#

Step 5: Defining the properties#

Loading an Entity with DLite#

1. Using the json storage plugin:#

2. Adding the file to storage_path#

Connect loaded entity to data#

Working with physical quantities#

2. Adding the file to `storage_path`#