Slide 1: Title Slide

Title:

Analysing and Modelling the Impact of Aggressive Driving on Crash Accident Duration
Using Telematics Data

Presenter Name

Department / University

Date

Slide 2:

Abstract – Overview

Study explores aggressive driving behaviours: overspeeding, harsh braking, and erratic
steering

Data from 34 long-distance public buses in Ethiopia

Utilized 6 months of pre-crash telematics data

Crash reports sourced from PSTS Insurance Department

Slide 3:

Abstract – Methodology

Tools: Jamovi v2.4.5, JASP v0.18.1

Key analyses:

Descriptive Statistics & Correlation

Cox Proportional Hazards Regression

Kaplan–Meier Survival Curves

Slide 4:

Abstract – Findings

Frequent overspeeding and harsh braking increased crash risk

Erratic steering also had moderate risk contribution

Telematics data enabled predictive modelling of crash timing

---

Slide 5:

Abstract – Conclusion & Recommendation

Early identification of high-risk drivers is possible

Authorities should adopt telematics-based monitoring

Supports data-driven training and policy design

---

Chapter One Slides

---
Slide 6:

Introduction – Road Safety in LMICs

Road traffic crashes = global public health issue

1.35 million deaths/year globally (WHO, 2022)

Over 90% of deaths occur in LMICs

Behavioural factors play a major role

---

Slide 7:

Role of Telematics in Safety Analytics

Telematics = GPS + sensor-based monitoring

Tracks speed, braking, lane changes

Enables real-time driver performance analysis

Shift from reactive to predictive safety management

---

Slide 8:

Aggressive Driving Behaviours

Includes: Overspeeding, Harsh Braking, Lane Departure

Associated with loss of control, increased severity

Global evidence links these to higher crash rates

---

Slide 9:

Global Context of Road Safety

Fatalities: 1.35M globally; mostly LMICs

Countries adopting data-centric safety tools

Use of machine learning and survival models is growing

---

Slide 10:

Ethiopian Road Safety Landscape

81% of crashes = driver error (EFTA, 2022)

$290 million lost/year

Urban growth worsens traffic & infrastructure strain

Lack of predictive monitoring tools

Slide 11:

Local Research Gaps

Few studies on when crashes happen, not just why

Lack of use of real-world telematics data

Existing crash analysis = retrospective, not predictive

---

Slide 12:

Rise of Telematics in Ethiopia

PSTS integrates GPS, accelerometer, gyroscope systems

Still underutilised in safety decision-making

Can bridge the gap in driver behaviour monitoring

---

Slide 13:

Statement of the Problem

Aggressive driving reduces crash-free time

No early-warning system currently exists

Critical to build time-to-event prediction models

---

Slide 14:

Justification of the Problem

Telematics = feasible tech for LMICs

Enables real-time risk assessment

Predictive analytics can save lives and reduce costs

Slide 15: General Objective

Use survival modelling (Cox, Kaplan-Meier) to estimate time-to-crash from telematics-

based behaviour scores

Slide 16:

Specific Objectives

Analyse overspeeding, harsh braking, and lane departure

Develop a composite aggressive driving score

Apply Cox PH to predict crash duration

---
Slide 17:

Research Questions

How does overspeeding affect crash timing?

What is the impact of harsh braking?

Can driving patterns predict when a crash might happen?

---

Slide 18:

Hypothesis

Aggressive driving significantly shortens time-to-crash

Strong positive correlation between event frequency and crash timing

---

Slide 19: Significance of the Study

Academic: introduces survival modelling in road safety

Policy: supports data-based intervention

Fleet: early risk detection, informed retraining

Insurance: risk-based driver profiling

---

Slide 20:

Scope and Limitations

Focus on 34 buses, 3 aggressive parameters

Limited by:

Sensor quality

Lack of environmental/contextual variables

Small sample but rich behavioural data

Sham A, [5/25/2025 1:08 AM]

Improved PowerPoint Slide Outline: Abstract + Chapter 1

With 3-4 clear and informative bullet points per slide as requested

Slide 1: Title Slide

Title: Modelling Crash Occurrence Using Telematics-Based Aggressive Driving Indicators

Subtitle: A Time-to-Crash Predictive Approach Using Cox Regression and Kaplan-Meier

Analysis

Your Name, Institution, Year

Slide 2: Abstract (1/2)

Aggressive driving like overspeeding, harsh braking, and erratic steering are major crash
contributors.

Study scope: Uses 6-month pre-crash telematics data from 34 public long-distance buses.

Tools: Cox Proportional Hazards Model and Kaplan-Meier estimators were used to model
time-to-crash.

Data Sources: Crash records from PSTS Insurance Department and telematics data logs.

Slide 3: Abstract (2/2)

Findings: Overspeeding and harsh braking increase crash likelihood significantly.

Telematics: Offers effective real-time monitoring and predictive analytics for fleet safety.

Policy Recommendation: Urges transport authorities to use telematics for high-risk driver
monitoring.

Key Terms: Telematics, survival analysis, risk modelling, Cox regression, aggressive
driving.

Slide 4: Chapter 1 – Introduction

Context: Traffic crashes are a global health and economic issue, especially in LMICs.

Statistics: WHO estimates 1.35 million deaths/year globally due to road crashes.

Problem: Ethiopia has rising crash rates due to driver-related aggressive behaviours.

Role of Telematics: Enables proactive crash prediction using real-time data.

Slide 5: Global Road Safety Context

Burden: 90% of traffic deaths occur in LMICs despite lower vehicle ownership.
Aggressive driving: Key cause of road crashes globally.

Technological Interventions: Predictive analytics and telematics help reduce crash rates.

Data Need: Real-time behaviour data improves response and prevention.

Slide 6: Ethiopia’s Road Safety Landscape

Crash Contribution: Over 80% of crashes caused by human error in Ethiopia.

Challenges: Weak enforcement, infrastructure limits, lack of telematics use.

Losses: $290M annual loss (~0.9% of GDP) from traffic crashes.

Solution Needed: Evidence-based behavioural interventions.

Slide 7: Rise of Telematics in Road Safety

Telematics Features: Tracks speed, braking, steering using GPS, gyroscopes,

accelerometers.

Application: Used in driver coaching, insurance pricing, crash prevention.

Global Adoption: Over 12 million systems in use globally.

Local Potential: Can support public transport safety in Ethiopia.

Slide 8: Statement of the Problem

Crash risk is behaviour-linked: OS, HB, and LD cause early crashes.

Monitoring Gap: Ethiopia lacks proactive monitoring tools.

Missed Opportunity: No national integration of real-time driver risk scoring.

Research Need: Modelling when crashes happen, not just if they do.

Slide 9: Problem Justification

Reactive Safety System: Ethiopia's current approach is backward-looking.

Aggressive Driving: Speeds up crash timing, often unnoticed.

Data Limitation: Reports are post-crash; few studies use pre-crash telematics.

Study Aim: Turn behaviour logs into predictive tools.

Slide 10: General & Specific Objectives

General Objective: Model time-to-crash using aggressive driving indicators.

Specific Objectives:
Analyse overspeeding behaviour per 100km.

Evaluate harsh braking frequency.

Examine erratic steering/lane departure.

Build a predictive model from composite scores.

Slide 11: Research Questions & Hypothesis

Questions:

How does overspeeding affect crash timing?

What is the crash impact of harsh braking?

Can erratic steering predict crash likelihood?

Hypothesis: Aggressive driving shortens crash-free time.

Slide 12: Significance of the Study

Academic Contribution: Introduces time-based crash modelling in Ethiopia.

Policy Impact: Supports data-driven road safety strategies.

Sham A, [5/25/2025 1:08 AM]

Fleet Safety: Enables identification of high-risk drivers.

Tech Innovation: Encourages adoption of telematics in Africa.

Slide 13: Scope & Limitations

Scope:

34 buses

3 behaviours: OS, HB, LD

Time-to-crash analysis

Limitations:

No data on fatigue or mobile use

Small sample size

Limited post-crash behaviour data

Slide 14: Delimitations of the Study

Focus: Single driver per bus, controlled variable.

Variables: OS, HB, and LD only.

Data Sources: PSTS insurance + telematics departments.

Crash Timing: Not crash severity.

Let me know if you'd like this exported to PowerPoint format or if you're ready for Chapter
2 slides. We can also begin compiling all of this into a single presentation file for
submission or defense.

Sham A, [5/25/2025 1:09 AM]

Chapter 2 PowerPoint Slide Outline: Literature Review

---

Slide 1: Chapter 2 - Literature Review

Focuses on key concepts, theoretical frameworks, and modelling techniques

Reviews global, regional, and local research findings

Guides the methodological choices of the current study

---

Slide 2: 2.1 Introduction to Aggressive Driving

Aggressive driving includes overspeeding, harsh braking, erratic steering

A global road safety concern causing over 1.35 million deaths annually (WHO, 2022)

Ethiopian context: high crash rates with minimal real-time behavioural monitoring

---

Slide 3: 2.2 Conceptual & Operational Definitions

Clarifies terms for consistency and analysis

Crash Time: Time from telematics monitoring start to crash event

Aggressive Driving: Any behaviour increasing crash risk (Shinar, 2017)

Telematics Data: Real-time sensor data (speed, braking, steering)

---

Slide 4: 2.2.1–2.2.8 Key Definitions

Crash Time: Modelled using survival analysis (Chen et al., 2020)

Overspeeding Index, Harsh Braking Rate, Lane Departure Index

Composite Risk Score (ADS)

Predictive Modelling: Statistical/ML tools to forecast crash time

---

Slide 5: 2.3 Theoretical Framework Overview

Three key theories:

1. Behavioural Theory

2. Risk Homeostasis Theory

3. Accident Causation Theory

Explain driver behaviour and crash causality using telematics insights

---

Slide 6: Behavioural Theory

Risky behaviours reinforced by immediate rewards (Skinner, 1953)

Telematics measures frequency, duration of risky events

Feedback mechanisms (dashboards, penalties) reduce risky habits

---

Slide 7: Risk Homeostasis Theory

Drivers adjust behaviour to perceived risk levels (Wilde, 1982)

Telematics can affect perception by real-time alerts or penalties

Important in public fleets where external stressors exist

---

Slide 8: Accident Causation Theory

Unsafe behaviours trigger crash sequence (Reason, 1990)

Modern models: Swiss Cheese framework

Useful in telematics: tracks precursors (speeding, erratic control)

---

Slide 9: 2.4 Empirical Literature Review - Global Studies

Studies show speeding and harsh braking linked to higher crash risk

Guo et al. (2020), Zhang et al. (2019), Dingus et al. (2016)

Telematics used to identify high-risk behaviours 30-90 days before crashes

---

Slide 10: Empirical Evidence from LMICs

Infrastructure gaps and policy enforcement limit effectiveness

Chigozie et al. (2021) Nigeria, Reddy & Sharma (2020) India

Risk scores & behaviour monitoring reduce accidents

---

Slide 11: Empirical Evidence from Ethiopia

76%-81% of crashes due to aggressive driving (Tulu et al., 2022)

Telematics usage growing, but limited

Public transport crash patterns validated with driver behaviour data

---

Slide 12: 2.5 Aggressive Driving and Crash Risk

Overspeeding: WHO (2023): 1% speed rise = 4% higher crash risk

Harsh Braking: Eboli et al. (2020) linked to 2.2x crash risk

Erratic Steering: Zhang et al. (2022) increased side-swipe likelihood

---

Slide 13: 2.6 Predictive Modelling with Telematics

Logistic Regression lacks time dimension

Cox PH Model & KM Estimation handle censored/time-based data

Feature engineering (events per 100 km) improves accuracy

---

Slide 14: 2.7 Research Gaps in Ethiopia

Lack of integrated telematics + crash data

Few studies use survival analysis

Limited feature engineering and ML applications

Sample size limits ML but supports Cox model

---

Slide 15: 2.8 Summary of Key Insights

AD behaviours are measurable and predictive

Telematics + survival analysis = better crash modelling

This study fills a local research gap and supports proactive safety policies

---

Next Step Shall we proceed to slide visuals or Chapter 3 outline?

Sham A, [5/25/2025 1:44 AM]

Great — now that I have the full version of Chapter 3: Materials and Methodology, I’ll
prepare your PowerPoint slides following your preferences:

Formal and clear structure

3–5 bullets per slide, with short, clear explanations

Balanced detail: not too short, not overloaded

No images unless you later approve

Here is your Chapter 3 PPT draft, slide-by-slide:

---

Slide 1: Chapter Three – Materials and Methodology

Focuses on modelling the time to crash using driver behaviour.

Applies a quantitative predictive approach based on survival analysis.

Uses aggressive driving indicators: overspeeding, harsh braking, erratic steering.

Covers 34 long-distance PSTS buses in Ethiopia.

---

Slide 2: 3.1 Study Approach

Quantitative method with predictive modelling.

Focus: time-to-event analysis (weeks/days to crash).

Based on survival modelling (Cox PH, Kaplan-Meier).

Objective: Link aggressive driving to crash timing.

---

Slide 3: 3.2 Target Population and Sampling

Target: Cross-country PSTS buses, not city buses.

Sample: 34 buses selected with consistent single drivers.

Selection based on crash involvement linked to aggressive driving.

City buses excluded due to traffic-based low-risk conditions.

---

Slide 4: 3.2.2 Inclusion Criteria

Each vehicle driven by one driver for 6+ months pre-crash.

Crash records show PSTS driver as primary cause.

Only buses with valid, complete telematics data were included.

Sample filtered from PSTS insurance & telematics data.

---

Slide 5: 3.2.3 Sampling Method and Timeframe

Purposive sampling based on availability and relevance.

Based on Cochran’s formula with 90% confidence, 10% margin.

6-month historical data window before each crash.

Ensures strong behavioural patterns before incident.

---

Slide 6: 3.3 Rationale for Cross-Country Buses

Higher speeds & more exposure to aggressive driving.

Greater risk of severe crashes due to momentum.

Urban crashes: low-speed, less predictive power.

Cross-country: clearer behavioural patterns for modelling.

---

Slide 7: Crash Severity vs Momentum (Concept)

Cross-country crashes = higher momentum = greater impact.

Conservation of momentum applied to understand force.

Visual difference: high-speed vs low-speed collisions.

Validates need to focus on high-speed (intercity) buses.

---

Slide 8: 3.4 Tools and Techniques

Survival analysis preferred over logistic regression.

Cox PH: links behaviour to crash hazard rate.

Kaplan-Meier: compares survival time between risk groups.

Logistic & ML models excluded for lack of time-based focus or interpretability.

---

Slide 9: 3.4.2 Data Sources and Platforms

Crash records from PSTS Insurance Department.

Telematics logs from PSTS Telematics Department.

Variables: speed, braking, steering, GPS, timestamps.

Data merged and filtered by inclusion rules.

---

Slide 10: 3.4.3 Analysis Software Used

Jamovi v2.4.5: descriptive stats, correlation, survival models.

JASP v0.18.1: cross-validation and visualisation.

Excel 2016: data cleaning, feature engineering, preliminary plots.

---

Slide 11: 3.4.4 Feature Engineering Overview

Events normalized per 100 km for comparability.

Metrics computed:
Overspeeding time/sec per 100 km

Harsh braking per 100 km

Lane departures per 100 km

Created a Composite Risk Score (CRS) from all metrics.

---

Slide 12: 3.5 Study Procedure Overview

9 structured steps from crash ID to modelling.

1. Identify crash vehicles

2. Retrieve telematics

3. Clean & pre-process data

4. Feature engineering

5. Compute time-to-crash

6. Perform EDA

7. Run survival models

8. Group risk levels

9. Report results

---

Slide 13: 3.5 Feature Engineering Calculations

OSI = (Overspeeding time ÷ distance) × 100

HBF = (Harsh braking events ÷ distance) × 100

LDR = (Lane departures ÷ distance) × 100

CADS = 0.36OSI + 0.33HBF + 0.32LDR

---

Slide 14: 3.5 Time-to-Crash Modelling

Crash duration calculated in weeks.

Vehicles without full data were censored.

Survival time used as dependent variable.

Sham A, [5/25/2025 1:44 AM]

Independent variables: OSI, HBF, LDR, CRS.

---

Slide 15: 3.6 Sample Size and Justification

Sample size = 34, based on Cochran’s formula.

Reflects real-world data limits with valid completeness.

Supports analytical strength of Cox regression.

---

Slide 16: 3.6 Feature Table (Overview)

Feature | Metric | Explanation

Overspeeding | Time & frequency per 100 km | Duration above speed limit

Harsh Braking | Events per 100 km | Sudden deceleration

Lane Departure | Events per 100 km | Erratic steering or veering

---

Slide 17: 3.7 Scoring Process Overview

Raw data → structured risk scores

Used formulas in Excel and Jamovi

Composite scoring weights:

0.36 (OS), 0.33 (HB), 0.32 (LD)

Risk categories:

Low: <0.33

Moderate: 0.33–0.66

High: ≥0.66

---

Slide 18: 3.8 Dataset Structure and Naming

Dataset rows = one vehicle, pre-crash metrics

Columns include:

ID, Distance, Crash date, OS, HB, LD, Risk score

Saved in Excel, analysed in Jamovi/JASP

Colour-coded by crash risk (Red, Yellow, Green)

---

Slide 19: 3.9 Ethical Considerations

Anonymised Data: Driver IDs masked

Data Security: Stored securely, access controlled

Non-Punitive Purpose: Findings used for safety, not punishment

Research ethics followed in full compliance.

---

Would you like this exported to a PowerPoint (.pptx) file next? Or should I begin Chapter
4 PPT prep now?

Sham A, [5/25/2025 2:14 AM]

Excellent. Now that you've provided the full Chapter 4 content and since we're temporarily
unable to generate visual slides due to plan limits, I will go ahead and prepare the full
PowerPoint-style slide content for Chapter 4 — in your preferred format with 3–5
meaningful, explained bullet points per slide.
---

CHAPTER 4: RESULTS AND DISCUSSIONS – POWERPOINT CONTENT

---

Slide 1: Chapter 4 – Results and Discussions

Presents analysis from six months of telematics data on 34 public buses.

Focus on aggressive driving behaviours: overspeeding, harsh braking, and erratic steering.

Statistical tools used: descriptive statistics, correlation, Cox regression, Kaplan–Meier

survival curves.

Aims to model time-to-crash and categorise driver risk.

---

Slide 2: 4.2 Descriptive & Exploratory Data Analysis

Most drivers showed moderate driving patterns; a few showed extreme aggression.
Overspeeding was more common than harsh braking or lane departures.

Histogram and boxplots revealed outliers with excessive risky behaviour.

Variability in driving patterns helped define high vs. low-risk profiles.

---

Slide 3: Descriptive Statistics Summary (Table 3)

Mean OS Duration = 135.7 sec per 100 km (max = 725 sec).

Harsh Braking less frequent but consistent indicator (avg. 1.9/100 km).

Lane Departure averaged 2.6 per 100 km – minor but still relevant.

Indicates overspeeding is most prevalent aggressive behaviour.

---

Slide 4: Sample Feature Extraction (Table 4)

Data normalised per 100 km to allow fair comparisons.

Vehicle 100AC had highest overspeeding duration: 15.37 sec/100 km.

Vehicle 100AD showed high harsh braking: 0.63/100 km.

Values illustrate variation across individual drivers.

---

Slide 5: 4.3 Cox Proportional Hazards Modelling

Time-to-event model used to estimate crash risk over time.

Chosen for flexibility and handling of censored data.

Overspeeding, harsh braking, lane departure included as covariates.

Assessed influence of each on time to crash.

---

Slide 6: Cox Regression Output (Table 5)

Overspeeding: HR = 1.16, statistically significant (p = 0.012).

Harsh Braking: HR = 1.07, moderate but significant (p = 0.038).

Lane Departure: HR = 1.03, weak and not statistically significant.

Confirms speeding as the most critical crash predictor.

---

Slide 7: Kaplan–Meier Survival Curve Analysis

Curves show how long drivers stayed crash-free.

Low Risk (Green): >2 years crash-free for most drivers.

Moderate Risk (Yellow): Risk increases sharply after 65 weeks.

High Risk (Red): Sharp crash risk rise within 30 weeks.

Helps visualise behavioural impact on crash timing.

---
Slide 8: Correlation Analysis (Table 6)

Overspeeding vs Time to Crash: r = -0.67 (strong negative).

Harsh braking and lane departures also showed negative correlation.

Risky behaviours were also interrelated, suggesting a consistent driving style.

Supports use of a composite score for driver risk.

---

Slide 9: Model Findings – Hazard Ratios

Every 10 seconds of overspeeding increases crash risk by 16%.

Each harsh braking event adds 7% risk.

Each lane departure adds 3%, marginally.

Emphasises need for overspeeding control.

---

Slide 10: Risk Scoring & Driver Categorisation

Composite Aggressive Driving Score (ADS) derived using HR weights.

Low Risk: ADS ≤ 0.40 → Crash-free up to 2.5 years.

Moderate Risk: ADS 0.41–0.69 → Crash likely within 1–1.5 years.

High Risk: ADS ≥ 0.70 → Crash expected in 6–12 months.

Enables targeted training, monitoring, and intervention.

---

Slide 11: Visualization of Survival Probabilities

Kaplan–Meier survival curves show declining crash-free probability over time.

Distinct separation between risk groups adds model confidence.

Visual tools help managers understand which drivers need immediate action.

Supports practical use of telematics data for prevention.

---

Slide 12: Survival Model Diagnostics

Model passed all diagnostic checks (VIF < 5, no multicollinearity).

Proportional Hazards assumption met (verified by Schoenfeld plots).

Sham A, [5/25/2025 2:14 AM]

No time-dependent effects found.

Cox model confirmed as valid and reliable for this telematics data.

---

Slide 13: Summary of Key Findings

Overspeeding is the strongest predictor of earlier crash time.

All three behaviours (OS, HB, LD) together shorten crash-free driving time.

Composite risk scoring can effectively classify drivers.

Public fleets can apply this model to reduce crash incidents and improve monitoring.

---

Would you like me to save this full slide deck draft as a Word or PowerPoint file once your
image generation limit resets? Or should we continue with Chapter 5: Conclusions and
Recommendations next?

Sham A, [5/25/2025 2:40 AM]

Here is your PowerPoint-ready content for the Conclusions and Recommendations section,
structured to match your preferred presentation style (with short but clear bullet points and
brief explanations per slide):

---

Slide 1: Chapter 5 – Conclusions & Recommendations

Objective Recap:

To analyse the time until crash based on aggressive driving behaviours (overspeeding,
harsh braking, erratic steering).

Method Applied:

Used Cox Regression & Kaplan–Meier survival models on telematics data from 34
Ethiopian public buses.
Goal:

Develop predictive models to identify crash timing and support early driver intervention.

---

Slide 2: Key Conclusion – Crash Predictability

Overspeeding is the top risk:

HR = 1.16 → 16% increased crash risk per 10s overspeeding/100 km.

Other contributors:

Harsh Braking (HR = 1.07)

Lane Departure (HR = 1.03)

These increase risk cumulatively.

Drivers with higher aggressive scores crash earlier.

---

Slide 3: Survival Analysis Insights

Low-Risk Drivers:

Stayed crash-free for over 2 years.

Moderate-Risk Drivers:

Crashes likely after 65 weeks.

High-Risk Drivers:

Crashed within 6 to 12 months.

Telematics signals are strong predictors of crash timing.

---

Slide 4: Practical Recommendations – Fleet Operators

Real-Time Monitoring:

Track overspeeding, HB, and LD using study thresholds.

Driver Risk Profiling:

Use composite risk scores to detect high-risk drivers.

Trigger-based Feedback:

Provide immediate training/interventions based on aggressive event thresholds.

---

Slide 5: Policy-Level Recommendations

Integrate Telematics in Fleet Regulation:

Mandatory in all government and private public fleets.

Set Risk Thresholds:

Enforce behaviour limits (e.g., speed >10/100 km flagged).

Retraining for High-Risk Drivers:

Base on Cox model & survival curve outcomes.

Support Predictive Analytics Platforms:

For crash prevention and insurance calibration.

---

Slide 6: Academic & Research Suggestions

Add More Behavioural Parameters:

Include distraction, phone use, fatigue in future models.

Explore Advanced Models:

Deep learning and Bayesian survival for higher accuracy.

Expand Scope:

Urban buses, motorcycles, light vehicles—wider relevance.

---

Slide 7: Limitations and Future Directions

Sample Size Constraint:

Only 34 buses – limits generalisability.

Context Bias:

Urban congestion patterns not fully captured.

Data Gaps:

No inclusion of crash severity, road type, or weather conditions.

Future Work:

Broader datasets + additional risk variables = richer models.

---
Slide 8: Final Thought

Predictive Safety is Achievable:

Using telematics to model crash timing saves lives.

From Reactive to Proactive Safety:

Intervene before the crash, not after.

Policy + Technology + Research = Safer Roads in LMICs

---

Would you like this exported into a PowerPoint (.pptx) file as well when your image
generation limit resets?