mpvis 1.0.5

Package for Multi-Perspective Process Visualization

mpvis

Multi-Perspective Process Visualization for Event Logs

mpvis is a Python library for discovering and visualizing business process models from event logs. It provides powerful tools to create multi-perspective Directly-Follows Graphs (DFG) and multi-dimensional Directed-Rooted Trees (DRT), enabling comprehensive process analysis across multiple dimensions including time, cost, quality, and flexibility.

Table of Contents

• Features
• Installation
• Quick Start
• API Reference
  • Event Log Formatting
  • Event Log Preprocessing
  • Multi-Perspective Directly-Follows Graph (MPDFG)
  • Multi-Dimensional Directed-Rooted Tree (MDDRT)
• Examples
• Requirements

Features

• Event Log Formatting: Standardize your event logs to the pm4py format
• Log Preprocessing: Filter and group activities for better visualization
• Multi-Perspective DFG: Discover process flows with frequency, time, and cost perspectives
• Multi-Dimensional DRT: Visualize process trees with quality and flexibility metrics
• Flexible Visualization: Export diagrams in multiple formats (SVG, PNG, PDF, etc.)
• Interactive Support: View diagrams directly in Jupyter Notebooks and Google Colab

Installation

mpvis requires Python 3.9 or higher. Install using pip:

pip install mpvis

Additional Requirements

To render and save generated diagrams, you must install Graphviz:

macOS (using Homebrew):

brew install graphviz

Ubuntu/Debian:

sudo apt-get install graphviz

Windows: Download and install from graphviz.org

Quick Start

import mpvis
import pandas as pd

# Load your event log
raw_log = pd.read_csv("event_log.csv")

# Format the log
format_dict = {
    "case:concept:name": "Case ID",
    "concept:name": "Activity",
    "time:timestamp": "Timestamp",
    "start_timestamp": "",  # Optional
    "org:resource": "",     # Optional
    "cost:total": ""        # Optional
}
event_log = mpvis.log_formatter(raw_log, format_dict)

# Discover a Multi-Perspective DFG
from mpvis import mpdfg

dfg, start_activities, end_activities = mpdfg.discover_multi_perspective_dfg(
    event_log,
    calculate_frequency=True,
    calculate_time=True,
    calculate_cost=True
)

# Visualize the DFG
mpdfg.view_multi_perspective_dfg(
    dfg,
    start_activities,
    end_activities,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True
)

API Reference

Event Log Formatting

log_formatter()

Formats your event log to comply with the pm4py standard column naming convention.

mpvis.log_formatter(log, log_format, timestamp_format=None)

Parameters:

• log (pd.DataFrame): The raw event log DataFrame
• log_format (dict): Dictionary mapping standard column names to your log's column names
• timestamp_format (str, optional): Format string for parsing timestamps (e.g., "%Y-%m-%d %H:%M:%S")

Format Dictionary Structure:

{
    "case:concept:name": "<Your Case ID Column>",      # Required
    "concept:name": "<Your Activity Column>",          # Required
    "time:timestamp": "<Your Timestamp Column>",       # Required
    "start_timestamp": "<Your Start Time Column>",     # Optional (use "" if not available)
    "org:resource": "<Your Resource Column>",          # Optional (use "" if not available)
    "cost:total": "<Your Cost Column>"                 # Optional (use "" if not available)
}

Returns:

• pd.DataFrame: Formatted event log with standardized column names

Example:

import mpvis
import pandas as pd

raw_log = pd.read_csv("process_data.csv")

format_dict = {
    "case:concept:name": "Case ID",
    "concept:name": "Activity Name",
    "time:timestamp": "Completion Time",
    "start_timestamp": "Start Time",
    "org:resource": "Resource",
    "cost:total": "Total Cost"
}

formatted_log = mpvis.log_formatter(raw_log, format_dict)

---

Event Log Preprocessing

manual_log_grouping()

Groups multiple specified activities into a single grouped activity within the event log.

from mpvis import preprocessing

preprocessing.manual_log_grouping(
    log,
    activities_to_group,
    group_name=None,
    case_id_key="case:concept:name",
    activity_id_key="concept:name",
    start_timestamp_key="start_timestamp",
    timestamp_key="time:timestamp"
)

Parameters:

• log (pd.DataFrame): The event log to process
• activities_to_group (list[str]): List of activity names to group together
• group_name (str, optional): Name for the grouped activity. Defaults to a concatenation of activity names
• case_id_key (str, optional): Column name for case IDs. Defaults to "case:concept:name"
• activity_id_key (str, optional): Column name for activity names. Defaults to "concept:name"
• start_timestamp_key (str, optional): Column name for start timestamps. Defaults to "start_timestamp"
• timestamp_key (str, optional): Column name for timestamps. Defaults to "time:timestamp"

Returns:

• pd.DataFrame: Event log with specified activities grouped

Example:

from mpvis import preprocessing

# Group related activities
activities_to_group = ["Check Documents", "Verify Documents", "Approve Documents"]

grouped_log = preprocessing.manual_log_grouping(
    event_log=event_log,
    activities_to_group=activities_to_group,
    group_name="Document Processing"
)

---

prune_log_based_on_top_variants()

Filters the event log to retain only the most frequent process variants, reducing complexity while preserving the most common paths.

from mpvis import preprocessing

preprocessing.prune_log_based_on_top_variants(
    log,
    k,
    activity_key="concept:name",
    timestamp_key="time:timestamp",
    case_id_key="case:concept:name"
)

Parameters:

• log (pd.DataFrame): The event log to prune
• k (int): Number of top variants to retain
• activity_key (str, optional): Column name for activities. Defaults to "concept:name"
• timestamp_key (str, optional): Column name for timestamps. Defaults to "time:timestamp"
• case_id_key (str, optional): Column name for case IDs. Defaults to "case:concept:name"

Returns:

• pd.DataFrame: Pruned event log containing only the top k variants

Example:

from mpvis import preprocessing

# Keep only the top 10 most frequent variants
pruned_log = preprocessing.prune_log_based_on_top_variants(
    log=event_log,
    k=10
)

---

Multi-Perspective Directly-Follows Graph (MPDFG)

Multi-Perspective Directly-Follows Graphs visualize the flow of activities in a process with multiple performance metrics including frequency, time, and cost.

discover_multi_perspective_dfg()

Discovers a multi-perspective DFG from an event log, extracting process flow information with multiple performance dimensions.

from mpvis import mpdfg

mpdfg.discover_multi_perspective_dfg(
    log,
    case_id_key="case:concept:name",
    activity_key="concept:name",
    timestamp_key="time:timestamp",
    start_timestamp_key="start_timestamp",
    cost_key="cost:total",
    calculate_frequency=True,
    calculate_time=True,
    calculate_cost=True,
    frequency_statistic="absolute-activity",
    time_statistic="mean",
    cost_statistic="mean"
)

Parameters:

• log (pd.DataFrame): The event log
• case_id_key (str, optional): Column name for case IDs. Defaults to "case:concept:name"
• activity_key (str, optional): Column name for activities. Defaults to "concept:name"
• timestamp_key (str, optional): Column name for timestamps. Defaults to "time:timestamp"
• start_timestamp_key (str, optional): Column name for start timestamps. Defaults to "start_timestamp"
• cost_key (str, optional): Column name for costs. Defaults to "cost:total"
• calculate_frequency (bool, optional): Whether to calculate frequencies. Defaults to True
• calculate_time (bool, optional): Whether to calculate time metrics. Defaults to True
• calculate_cost (bool, optional): Whether to calculate cost metrics. Defaults to True
• frequency_statistic (str, optional): Frequency calculation method. Options: "absolute-activity", "absolute-case", "relative-activity", "relative-case". Defaults to "absolute-activity"
• time_statistic (str, optional): Time aggregation method. Options: "mean", "sum", "max", "min", "median", "stdev". Defaults to "mean"
• cost_statistic (str, optional): Cost aggregation method. Options: "mean", "sum", "max", "min", "median", "stdev". Defaults to "mean"

Returns:

• Tuple[dict, dict, dict]: A tuple containing:
  • Multi-perspective DFG dictionary
  • Start activities dictionary
  • End activities dictionary

Example:

from mpvis import mpdfg

dfg, start_activities, end_activities = mpdfg.discover_multi_perspective_dfg(
    event_log,
    calculate_frequency=True,
    calculate_time=True,
    calculate_cost=True,
    frequency_statistic="absolute-activity",
    time_statistic="median",
    cost_statistic="sum"
)

---

filter_multi_perspective_dfg_activities()

Filters activities in the DFG based on a specified metric, showing only a percentage of the most or least significant activities.

from mpvis import mpdfg

mpdfg.filter_multi_perspective_dfg_activities(
    percentage,
    multi_perspective_dfg,
    start_activities,
    end_activities,
    sort_by="frequency",
    ascending=True
)

Parameters:

• percentage (float): Percentage of activities to keep (0-100)
• multi_perspective_dfg (dict): The multi-perspective DFG
• start_activities (dict): Start activities dictionary
• end_activities (dict): End activities dictionary
• sort_by (str, optional): Metric to filter by. Options: "frequency", "time", "cost". Defaults to "frequency"
• ascending (bool, optional): If True, keeps activities with lowest values; if False, keeps highest. Defaults to True

Returns:

• dict: Filtered multi-perspective DFG

Example:

from mpvis import mpdfg

# Keep only the top 20% most frequent activities
filtered_dfg = mpdfg.filter_multi_perspective_dfg_activities(
    percentage=20,
    multi_perspective_dfg=dfg,
    start_activities=start_activities,
    end_activities=end_activities,
    sort_by="frequency",
    ascending=True
)

---

filter_multi_perspective_dfg_paths()

Filters paths (edges) in the DFG based on a specified metric.

from mpvis import mpdfg

mpdfg.filter_multi_perspective_dfg_paths(
    percentage,
    multi_perspective_dfg,
    start_activities,
    end_activities,
    sort_by="frequency",
    ascending=True
)

Parameters:

• percentage (float): Percentage of paths to keep (0-100)
• multi_perspective_dfg (dict): The multi-perspective DFG
• start_activities (dict): Start activities dictionary
• end_activities (dict): End activities dictionary
• sort_by (str, optional): Metric to filter by. Options: "frequency", "time". Defaults to "frequency"
• ascending (bool, optional): If True, keeps paths with lowest values; if False, keeps highest. Defaults to True

Returns:

• dict: Filtered multi-perspective DFG

Example:

from mpvis import mpdfg

# Keep only the top 30% most frequent paths
filtered_dfg = mpdfg.filter_multi_perspective_dfg_paths(
    percentage=30,
    multi_perspective_dfg=dfg,
    start_activities=start_activities,
    end_activities=end_activities,
    sort_by="frequency",
    ascending=True
)

---

get_multi_perspective_dfg_string()

Generates a string representation of the multi-perspective DFG diagram for programmatic use.

from mpvis import mpdfg

mpdfg.get_multi_perspective_dfg_string(
    multi_perspective_dfg,
    start_activities,
    end_activities,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True,
    cost_currency="USD",
    rankdir="TD",
    diagram_tool="graphviz",
    arc_thickness_by="frequency"
)

Parameters:

• multi_perspective_dfg (dict): The multi-perspective DFG
• start_activities (dict): Start activities dictionary
• end_activities (dict): End activities dictionary
• visualize_frequency (bool, optional): Show frequency metrics. Defaults to True
• visualize_time (bool, optional): Show time metrics. Defaults to True
• visualize_cost (bool, optional): Show cost metrics. Defaults to True
• cost_currency (str, optional): Currency symbol for costs. Defaults to "USD"
• rankdir (str, optional): Graph direction. Options: "TD" (top-down), "BT" (bottom-top), "LR" (left-right), "RL" (right-left). Defaults to "TD"
• diagram_tool (str, optional): Diagramming tool. Options: "graphviz", "mermaid". Defaults to "graphviz"
• arc_thickness_by (str, optional): Metric for edge thickness. Options: "frequency", "time". Defaults to "frequency"

Returns:

• str: String representation of the DFG diagram

Note:

• Mermaid diagrams can only be saved as HTML files and require internet connection to display properly

Example:

from mpvis import mpdfg

dfg_string = mpdfg.get_multi_perspective_dfg_string(
    dfg,
    start_activities,
    end_activities,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=False,
    rankdir="LR",
    diagram_tool="graphviz"
)

---

view_multi_perspective_dfg()

Displays the multi-perspective DFG in interactive environments like Jupyter Notebook or Google Colab.

from mpvis import mpdfg

mpdfg.view_multi_perspective_dfg(
    multi_perspective_dfg,
    start_activities,
    end_activities,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True,
    cost_currency="USD",
    rankdir="TD",
    format="svg",
    arc_thickness_by="frequency"
)

Parameters:

• multi_perspective_dfg (dict): The multi-perspective DFG
• start_activities (dict): Start activities dictionary
• end_activities (dict): End activities dictionary
• visualize_frequency (bool, optional): Show frequency metrics. Defaults to True
• visualize_time (bool, optional): Show time metrics. Defaults to True
• visualize_cost (bool, optional): Show cost metrics. Defaults to True
• cost_currency (str, optional): Currency symbol. Defaults to "USD"
• rankdir (str, optional): Graph direction. Defaults to "TD"
• format (str, optional): Image format. Options: "svg", "png", "jpg", "jpeg", "webp". Defaults to "svg"
• arc_thickness_by (str, optional): Metric for edge thickness. Options: "frequency", "time". Defaults to "frequency"

Returns:

• None (displays the diagram)

Note:

• Only supports Graphviz diagrams
• Not all formats are supported in all environments

Example:

from mpvis import mpdfg

mpdfg.view_multi_perspective_dfg(
    dfg,
    start_activities,
    end_activities,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True,
    rankdir="LR",
    format="svg"
)

---

save_vis_multi_perspective_dfg()

Saves the multi-perspective DFG visualization to a file.

from mpvis import mpdfg

mpdfg.save_vis_multi_perspective_dfg(
    multi_perspective_dfg,
    start_activities,
    end_activities,
    file_name,
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True,
    cost_currency="USD",
    format="svg",
    rankdir="TD",
    diagram_tool="graphviz",
    arc_thickness_by="frequency"
)

Parameters:

• multi_perspective_dfg (dict): The multi-perspective DFG
• start_activities (dict): Start activities dictionary
• end_activities (dict): End activities dictionary
• file_name (str): Output file path (without extension)
• visualize_frequency (bool, optional): Show frequency metrics. Defaults to True
• visualize_time (bool, optional): Show time metrics. Defaults to True
• visualize_cost (bool, optional): Show cost metrics. Defaults to True
• cost_currency (str, optional): Currency symbol. Defaults to "USD"
• format (str, optional): Output format. Options: "svg", "png", "pdf", "jpg", etc. See Graphviz outputs. Defaults to "svg"
• rankdir (str, optional): Graph direction. Defaults to "TD"
• diagram_tool (str, optional): Diagramming tool. Options: "graphviz", "mermaid". Defaults to "graphviz"
• arc_thickness_by (str, optional): Metric for edge thickness. Options: "frequency", "time". Defaults to "frequency"

Returns:

• None (saves the file)

Note:

• Mermaid diagrams are saved as HTML files only

Example:

from mpvis import mpdfg

mpdfg.save_vis_multi_perspective_dfg(
    dfg,
    start_activities,
    end_activities,
    file_name="process_diagram",
    visualize_frequency=True,
    visualize_time=True,
    visualize_cost=True,
    format="png",
    rankdir="LR"
)

---

Multi-Dimensional Directed-Rooted Tree (MDDRT)

Multi-Dimensional Directed-Rooted Trees provide hierarchical process visualizations with multiple performance dimensions including time, cost, quality, and flexibility.

discover_multi_dimensional_drt()

Discovers and constructs a multi-dimensional DRT from an event log.

from mpvis import mddrt

mddrt.discover_multi_dimensional_drt(
    log,
    calculate_time=True,
    calculate_cost=True,
    calculate_quality=True,
    calculate_flexibility=True,
    group_activities=False,
    show_names=False,
    case_id_key="case:concept:name",
    activity_key="concept:name",
    timestamp_key="time:timestamp",
    start_timestamp_key="start_timestamp",
    cost_key="cost:total"
)

Parameters:

• log (pd.DataFrame): The event log
• calculate_time (bool, optional): Include time dimension. Defaults to True
• calculate_cost (bool, optional): Include cost dimension. Defaults to True
• calculate_quality (bool, optional): Include quality dimension. Defaults to True
• calculate_flexibility (bool, optional): Include flexibility dimension. Defaults to True
• group_activities (bool, optional): Group sequential single-path activities. Defaults to False
• show_names (bool, optional): Show names of grouped activities. Defaults to False
• case_id_key (str, optional): Column name for case IDs. Defaults to "case:concept:name"
• activity_key (str, optional): Column name for activities. Defaults to "concept:name"
• timestamp_key (str, optional): Column name for timestamps. Defaults to "time:timestamp"
• start_timestamp_key (str, optional): Column name for start timestamps. Defaults to "start_timestamp"
• cost_key (str, optional): Column name for costs. Defaults to "cost:total"

Returns:

• TreeNode: Root node of the multi-dimensional DRT

Example:

from mpvis import mddrt

drt = mddrt.discover_multi_dimensional_drt(
    event_log,
    calculate_time=True,
    calculate_cost=True,
    calculate_quality=True,
    calculate_flexibility=True,
    group_activities=True,
    show_names=True
)

---

group_drt_activities()

Groups sequential activities that follow a single-child path in the DRT, simplifying the tree structure.

from mpvis import mddrt

mddrt.group_drt_activities(
    multi_dimensional_drt,
    show_names=False
)

Parameters:

• multi_dimensional_drt (TreeNode): Root of the multi-dimensional DRT
• show_names (bool, optional): Show names of grouped activities. Defaults to False

Returns:

• TreeNode: Root of the grouped multi-dimensional DRT

Example:

from mpvis import mddrt

# First discover the DRT
drt = mddrt.discover_multi_dimensional_drt(event_log)

# Then group activities
grouped_drt = mddrt.group_drt_activities(drt, show_names=True)

---

get_multi_dimensional_drt_string()

Generates a string representation of the multi-dimensional DRT diagram.

from mpvis import mddrt

mddrt.get_multi_dimensional_drt_string(
    multi_dimensional_drt,
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=True,
    visualize_flexibility=True,
    node_measures=["total"],
    arc_measures=[]
)

Parameters:

• multi_dimensional_drt (TreeNode): Root of the multi-dimensional DRT
• visualize_time (bool, optional): Include time dimension. Defaults to True
• visualize_cost (bool, optional): Include cost dimension. Defaults to True
• visualize_quality (bool, optional): Include quality dimension. Defaults to True
• visualize_flexibility (bool, optional): Include flexibility dimension. Defaults to True
• node_measures (list, optional): Node metrics to display. Options: "total", "consumed", "remaining". Defaults to ["total"]
• arc_measures (list, optional): Arc metrics to display. Options: "avg", "min", "max". Defaults to []

Returns:

• str: String representation of the DRT diagram

Example:

from mpvis import mddrt

drt_string = mddrt.get_multi_dimensional_drt_string(
    drt,
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=False,
    visualize_flexibility=False,
    node_measures=["total", "consumed"],
    arc_measures=["avg", "min", "max"]
)

---

view_multi_dimensional_drt()

Displays the multi-dimensional DRT in interactive environments.

from mpvis import mddrt

mddrt.view_multi_dimensional_drt(
    multi_dimensional_drt,
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=True,
    visualize_flexibility=True,
    node_measures=["total"],
    arc_measures=[],
    format="svg"
)

Parameters:

• multi_dimensional_drt (TreeNode): Root of the multi-dimensional DRT
• visualize_time (bool, optional): Include time dimension. Defaults to True
• visualize_cost (bool, optional): Include cost dimension. Defaults to True
• visualize_quality (bool, optional): Include quality dimension. Defaults to True
• visualize_flexibility (bool, optional): Include flexibility dimension. Defaults to True
• node_measures (list, optional): Node metrics to display. Options: "total", "consumed", "remaining". Defaults to ["total"]
• arc_measures (list, optional): Arc metrics to display. Options: "avg", "min", "max". Defaults to []
• format (str, optional): Image format. Options: "svg", "png", "jpg", "jpeg", "webp". Defaults to "svg"

Returns:

• None (displays the diagram)

Note:

• Not all output formats are supported in all interactive environments

Example:

from mpvis import mddrt

mddrt.view_multi_dimensional_drt(
    drt,
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=True,
    visualize_flexibility=True,
    node_measures=["total", "remaining"],
    arc_measures=["avg"],
    format="svg"
)

---

save_vis_multi_dimensional_drt()

Saves the multi-dimensional DRT visualization to a file.

from mpvis import mddrt

mddrt.save_vis_multi_dimensional_drt(
    multi_dimensional_drt,
    file_path,
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=True,
    visualize_flexibility=True,
    node_measures=["total"],
    arc_measures=[],
    format="svg"
)

Parameters:

• multi_dimensional_drt (TreeNode): Root of the multi-dimensional DRT
• file_path (str): Output file path (without extension)
• visualize_time (bool, optional): Include time dimension. Defaults to True
• visualize_cost (bool, optional): Include cost dimension. Defaults to True
• visualize_quality (bool, optional): Include quality dimension. Defaults to True
• visualize_flexibility (bool, optional): Include flexibility dimension. Defaults to True
• node_measures (list, optional): Node metrics to display. Options: "total", "consumed", "remaining". Defaults to ["total"]
• arc_measures (list, optional): Arc metrics to display. Options: "avg", "min", "max". Defaults to []
• format (str, optional): Output format. Options: "svg", "png", "pdf", "jpg", etc. Defaults to "svg"

Returns:

• None (saves the file)

Example:

from mpvis import mddrt

mddrt.save_vis_multi_dimensional_drt(
    drt,
    file_path="process_tree",
    visualize_time=True,
    visualize_cost=True,
    visualize_quality=True,
    visualize_flexibility=True,
    node_measures=["total", "consumed", "remaining"],
    arc_measures=["avg", "max"],
    format="png"
)

---

Examples

For comprehensive examples demonstrating mpvis capabilities with real-world event logs, check out the Examples directory in the GitHub repository.

The examples include:

• BPI Challenge datasets
• Hospital process analysis
• IT incident management
• Manufacturing processes
• And more...

Requirements

• Graphviz (system installation)

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.